# Terradev (remote · terradev-mcp.terradev.cloud)

Cross-cloud GPU orchestration CLI.

- Trust score: 63/100 (medium)
- Change this week: +3
- Registry status: active
- Liveness: live
- Owner verified: no
- Last scored: 2026-09-21

## Components

- remote · `terradev-mcp.terradev.cloud`: 63/100 (this document), [markdown](https://verifymcp.io/servers/theoddden-terradev/terradev-mcp.md), [page](https://verifymcp.io/servers/theoddden-terradev/terradev-mcp)
- pypi · `terradev-cli`: 76/100, [markdown](https://verifymcp.io/servers/theoddden-terradev/terradev-cli.md), [page](https://verifymcp.io/servers/theoddden-terradev/terradev-cli)

## Channel facts

- Endpoint: `https://terradev-mcp.terradev.cloud/sse`
- Transports: `sse`
- Auth: `none`
- Version: `6.2.20`

## Trust breakdown

How this component scores in each security and reliability category. Every signal is checked automatically against the live server, and we only credit what we can confirm. Scores are 0–100 per category. Scoring method: https://verifymcp.io/docs/scoring (what has changed: https://verifymcp.io/docs/scoring/changelog)

Scored 2026-09-21.

- **Endpoint Security**: 57/100
  - The endpoint's TLS certificate is valid, in date, and uses a strong key.
  - Authorisation not fully verified: no authorisation is required to call this server, and 245 tool(s) never declared a destructiveHint. The MCP spec treats an absent hint as destructive by default, so we cannot call this surface safe.
  - HTTPS is enforced; there's no plaintext access path.
  - HSTS check failed: the Strict-Transport-Security header is absent.
  - DNSSEC check failed: this domain isn't protected by DNSSEC.
- **Transport & Reachability**: 40/100
  - Verified sse transport via a live MCP handshake.
- **Schema Quality & AI Usability**: 79/100
  - 100% of prompts and resources have a non-trivial description (not blank, and not just the item's name).
  - AI-judged instruction clarity (good).
  - Context-footprint check failed: tool/resource definitions use about 27375 tokens (~109/item across 250 items; 245 tools + 5 resources), over budget; trim descriptions and params.
  - Usage-examples check failed: none of the tools include examples.
- **Stability & Change Management**: 40/100
  - Stability check failed: schema churn in the 13 days we've observed: 8 tool removals, 0 breaking changes, 0 auth/transport breaks, 8 additions.
- **Tool Coverage**: 100/100
  - 100% of tools have a non-trivial description (not blank, and not just the tool's name).
  - 100% of tool parameters carry a description.
- **Tool Safety**: 75/100
  - No prompt-injection markers were found in the server instructions, tool names or descriptions we captured.
  - 0 of 21 tool(s) whose name or description implies an irreversible operation declare an MCP destructiveHint annotation; "inferx_deploy" implies "deploy" and declares no destructiveHint at all, which the MCP spec reads as destructive by default.
  - An AI judge read all 246 captured unit(s) of tool text and found none that tries to manipulate the model reading it.
- **Capabilities**: 100/100
  - Implements a supported MCP spec version (2025-11-25); the latest is 2026-07-28.

## Install

### How do I install the Terradev MCP server?

Terradev is a hosted endpoint at https://terradev-mcp.terradev.cloud/sse, so there is nothing to install locally. Ready-made configuration for Claude, Cursor, VS Code, Codex and 5 more is on this page, copied from each client's own documentation.

### Claude

```bash
claude mcp add --transport http theoddden-terradev 'https://terradev-mcp.terradev.cloud/sse'
```

### Cursor

```json
{
  "mcpServers": {
    "theoddden-terradev": {
      "url": "https://terradev-mcp.terradev.cloud/sse"
    }
  }
}
```

### VS Code

```json
{
  "servers": {
    "theoddden-terradev": {
      "type": "http",
      "url": "https://terradev-mcp.terradev.cloud/sse"
    }
  }
}
```

### Codex

```toml
[mcp_servers.theoddden-terradev]
url = "https://terradev-mcp.terradev.cloud/sse"
```

### opencode

```json
{
  "$schema": "https://opencode.ai/config.json",
  "mcp": {
    "theoddden-terradev": {
      "type": "remote",
      "url": "https://terradev-mcp.terradev.cloud/sse",
      "enabled": true
    }
  }
}
```

### OpenClaw

```bash
openclaw mcp add theoddden-terradev --url 'https://terradev-mcp.terradev.cloud/sse' --transport streamable-http
```

### Hermes

```yaml
mcp_servers:
  theoddden-terradev:
    url: "https://terradev-mcp.terradev.cloud/sse"
```

### Netclaw

```json
{
  "McpServers": {
    "theoddden-terradev": {
      "Transport": "http",
      "Url": "https://terradev-mcp.terradev.cloud/sse"
    }
  }
}
```

### Vellum

```bash
assistant mcp add theoddden-terradev -t streamable-http -u 'https://terradev-mcp.terradev.cloud/sse'
```

### Other

```json
{
  "mcpServers": {
    "theoddden-terradev": {
      "type": "http",
      "url": "https://terradev-mcp.terradev.cloud/sse"
    }
  }
}
```

The mcpServers block is a cross-client convention. Remote transports vary, so check your client's docs.

## Changelog

Every change recorded for this component, newest first. Days that predate change tracking, or that we cannot explain, say so: "we were watching and nothing happened" and "we were not watching" are different claims.

### 2026-09-20 (score 63, +1)

No change was recorded against any check on this day. Stability & Change Management went from 33 to 37.

### 2026-09-18 (score 62, +1)

No change was recorded against any check on this day. Stability & Change Management went from 27 to 30.

### 2026-09-16 (score 61, +1)

No change was recorded against any check on this day. Stability & Change Management went from 20 to 23.

### 2026-09-14 (score 60, +1)

No change was recorded against any check on this day. Stability & Change Management went from 13 to 17.

### 2026-09-11 (score 59, +1)

No change was recorded against any check on this day. Stability & Change Management went from 3 to 7.

### 2026-09-10 (score 58, 0)

We did not load change detail this far back for this component, so this day may have recorded more than is shown.

- [security regression] Stability: 0.03 → fail
- [security regression] A breaking change shipped without a version bump: still 2.0.1
- [security regression] Tool “vault_run” was removed
- [security regression] Tool “vault_list” was removed
- [security regression] Tool “vault_remove” was removed
- [security regression] Tool “vault_env” was removed
- [security regression] Tool “vault_verify” was removed
- [security regression] Tool “vault_get” was removed
- [security regression] Tool “vault_set” was removed
- [security regression] Tool “vault_sync” was removed
- [functional] New tool “secrets_run”
- [functional] New tool “secrets_get”
- [functional] New tool “secrets_list”
- [functional] New tool “secrets_env”
- [functional] New tool “secrets_set”
- [functional] New tool “secrets_remove”
- [functional] New tool “secrets_verify”
- [functional] New tool “secrets_sync”
- [cosmetic] “langchain_create_sglang_pipeline” reworded the description of “credentials”
- [cosmetic] “phoenix_otel_env” reworded the description of “credentials”
- [cosmetic] “langfuse_score” reworded the description of “credentials”
- [cosmetic] “ray_status” reworded the description of “credentials”
- [cosmetic] “agent_deploy” reworded the description of “credentials”
- [cosmetic] “agent_mem0_get” reworded the description of “credentials”
- [cosmetic] “training_launch_distributed” reworded the description of “credentials”
- [cosmetic] “price_intel” reworded the description of “credentials”
- [cosmetic] “agent_mem0_update” reworded the description of “credentials”
- [cosmetic] “vllm_benchmark” reworded the description of “credentials”
- [cosmetic] “langchain_create_workflow” reworded the description of “credentials”
- [cosmetic] “train_unsloth_run” reworded the description of “credentials”
- [cosmetic] “run_workflow” reworded the description of “credentials”
- [cosmetic] “sglang_inference” reworded the description of “credentials”
- [cosmetic] “status” reworded the description of “credentials”
- [cosmetic] “active_context” reworded the description of “credentials”
- [cosmetic] “manage_instance” reworded the description of “credentials”
- [cosmetic] “warm_pool_start” reworded the description of “credentials”
- [cosmetic] “dvc_diff” reworded the description of “credentials”
- [cosmetic] “vllm_wake” reworded the description of “credentials”
- [cosmetic] “agent_status” reworded the description of “credentials”
- [cosmetic] “hf_smart_template” reworded the description of “credentials”
- [cosmetic] “provision_gpu” reworded the description of “credentials”
- [cosmetic] “train_snapshot” reworded the description of “credentials”
- [cosmetic] “checkpoint_delete” reworded the description of “credentials”
- [cosmetic] “hf_list_datasets” reworded the description of “credentials”
- [cosmetic] “database_weaviate_query” reworded the description of “credentials”
- [cosmetic] “sglang” reworded the description of “credentials”
- [cosmetic] “deepeval_metrics” reworded the description of “credentials”
- [cosmetic] “dvc_status” reworded the description of “credentials”
- [cosmetic] “dvc_push” reworded the description of “credentials”
- [cosmetic] “langgraph_workflow_status” reworded the description of “credentials”
- [cosmetic] “sglang_stop” reworded the description of “credentials”
- [cosmetic] “agent_langgraph_status” reworded the description of “credentials”
- [cosmetic] “phoenix_spans” reworded the description of “credentials”
- [cosmetic] “wandb_run_details” reworded the description of “credentials”
- [cosmetic] “phoenix_k8s” reworded the description of “credentials”
- [cosmetic] “guardrails_k8s” reworded the description of “credentials”
- [cosmetic] “langgraph_orchestrator_worker” reworded the description of “credentials”
- [cosmetic] “orchestrator_status” reworded the description of “credentials”
- [cosmetic] “langfuse_quality” reworded the description of “credentials”
- [cosmetic] “agent_mem0_add” reworded the description of “credentials”
- [cosmetic] “get_database_connection” reworded the description of “credentials”
- [cosmetic] “cost_simulate” reworded the description of “credentials”
- [cosmetic] “helm_generate” reworded the description of “credentials”
- [cosmetic] “agent_agentic_serving_show_config” reworded the description of “credentials”
- [cosmetic] “qdrant_collections” reworded the description of “credentials”
- [cosmetic] “agent_letta_list” reworded the description of “credentials”
- [cosmetic] “hf_create_endpoint” reworded the description of “credentials”
- [cosmetic] “inferx_usage” reworded the description of “credentials”
- [cosmetic] “ollama_model_info” reworded the description of “credentials”
- [cosmetic] “guardrails_chat” reworded the description of “credentials”
- [cosmetic] “train_stop” reworded the description of “credentials”
- [cosmetic] “infer_failover” reworded the description of “credentials”
- [cosmetic] “train_resume” reworded the description of “credentials”
- [cosmetic] “governance_request_consent” reworded the description of “credentials”
- [cosmetic] “ollama_pull” reworded the description of “credentials”
- [cosmetic] “dvc_stage_checkpoint” reworded the description of “credentials”
- [cosmetic] “agent_langchain_create_langgraph” reworded the description of “credentials”
- [cosmetic] “analytics” reworded the description of “credentials”
- [cosmetic] “langgraph_create_workflow” reworded the description of “credentials”
- [cosmetic] “sglang_start” reworded the description of “credentials”
- [cosmetic] “agent_cost” reworded the description of “credentials”
- [cosmetic] “agent_skill_attach” reworded the description of “credentials”
- [cosmetic] “preflight” reworded the description of “credentials”
- [cosmetic] “qdrant_info” reworded the description of “credentials”
- [cosmetic] “agent_teardown” reworded the description of “credentials”
- [cosmetic] “train_monitor” reworded the description of “credentials”
- [cosmetic] “egress_cheapest_route” reworded the description of “credentials”
- [cosmetic] “agent_agentic_serving_launch_args” reworded the description of “credentials”
- [cosmetic] “inferx_list” reworded the description of “credentials”
- [cosmetic] “database_weaviate_list_collections” reworded the description of “credentials”
- [cosmetic] “hf_delete_endpoint” reworded the description of “credentials”
- [cosmetic] “vllm_analyze” reworded the description of “credentials”
- [cosmetic] “hf_list_endpoints” reworded the description of “credentials”
- [cosmetic] “inferx_quote” reworded the description of “credentials”
- [cosmetic] “query_database” reworded the description of “credentials”
- [cosmetic] “ray_start” reworded the description of “credentials”
- [cosmetic] “price_trends” reworded the description of “credentials”
- [cosmetic] “checkpoint_list” reworded the description of “credentials”
- [cosmetic] “k8s_create” reworded the description of “credentials”
- [cosmetic] “hf_model_info” reworded the description of “credentials”
- [cosmetic] “preflight_report” reworded the description of “credentials”
- [cosmetic] “ollama_ps” reworded the description of “credentials”
- [cosmetic] “vllm_info” reworded the description of “credentials”
- [cosmetic] “inferx_optimize” reworded the description of “credentials”
- [cosmetic] “hf_space_deploy” reworded the description of “credentials”
- [cosmetic] “orchestrator_register” reworded the description of “credentials”
- [cosmetic] “lora_remove” reworded the description of “credentials”
- [cosmetic] “infer_route” reworded the description of “credentials”
- [cosmetic] “inferx_status” reworded the description of “credentials”
- [cosmetic] “hf_hardware_compare” reworded the description of “credentials”
- [cosmetic] “k8s_list” reworded the description of “credentials”
- [cosmetic] “checkpoint_save” reworded the description of “credentials”
- [cosmetic] “agent_letta_delete” reworded the description of “credentials”
- [cosmetic] “agent_langgraph_deploy” reworded the description of “credentials”
- [cosmetic] “agent_vector_db_up” reworded the description of “credentials”
- [cosmetic] “vllm_auto_optimize” reworded the description of “credentials”
- [cosmetic] “create_sqlite_connection” reworded the description of “credentials”
- [cosmetic] “train_status” reworded the description of “credentials”
- [cosmetic] “agent_skill_init” reworded the description of “credentials”
- [cosmetic] “langfuse_datasets” reworded the description of “credentials”
- [cosmetic] “governance_record_consent” reworded the description of “credentials”
- [cosmetic] “agent_letta_status” reworded the description of “credentials”
- [cosmetic] “agent_letta_chat” reworded the description of “credentials”
- [cosmetic] “database_weaviate_create_collection” reworded the description of “credentials”
- [cosmetic] “hf_endpoint_info” reworded the description of “credentials”
- [cosmetic] “ray_stop” reworded the description of “credentials”
- [cosmetic] “ml_vllm_lora_sync” reworded the description of “credentials”
- [cosmetic] “phoenix_snippet” reworded the description of “credentials”
- [cosmetic] “agent_langchain_test” reworded the description of “credentials”
- [cosmetic] “mlflow_log_run” reworded the description of “credentials”
- [cosmetic] “agent_langgraph_test” reworded the description of “credentials”
- [cosmetic] “egress_optimize_staging” reworded the description of “credentials”
- [cosmetic] “agent_mem0_search” reworded the description of “credentials”
- [cosmetic] “inferx_delete” reworded the description of “credentials”
- [cosmetic] “guardrails_generate_config” reworded the description of “credentials”
- [cosmetic] “k8s_device_plugin” reworded the description of “credentials”
- [cosmetic] “mlflow_register_model” reworded the description of “credentials”
- [cosmetic] “mlflow_list_experiments” reworded the description of “credentials”
- [cosmetic] “agent_mem0_forget” reworded the description of “credentials”
- [cosmetic] “k8s_gpu_operator_install” reworded the description of “credentials”
- [cosmetic] “agent_scale” reworded the description of “credentials”
- [cosmetic] “local_scan” reworded the description of “credentials”
- [cosmetic] “orchestrator_load” reworded the description of “credentials”
- [cosmetic] “deepeval_init” reworded the description of “credentials”
- [cosmetic] “langfuse_configure” reworded the description of “credentials”
- [cosmetic] “agent_langchain_create_workflow” reworded the description of “credentials”
- [cosmetic] “orchestrator_start” reworded the description of “credentials”
- [cosmetic] “agent_letta_create” reworded the description of “credentials”
- [cosmetic] “hf_space_status” reworded the description of “credentials”
- [cosmetic] “preflight_gpu_check” reworded the description of “credentials”
- [cosmetic] “k8s_time_slicing” reworded the description of “credentials”
- [cosmetic] “stage” reworded the description of “credentials”
- [cosmetic] “cost_optimize_recommend” reworded the description of “credentials”
- [cosmetic] “agent_agentic_serving_configure” reworded the description of “credentials”
- [cosmetic] “manifests” reworded the description of “credentials”
- [cosmetic] “optimize” reworded the description of “credentials”
- [cosmetic] “hf_list_models” reworded the description of “credentials”
- [cosmetic] “agent_agentic_serving_k8s” reworded the description of “credentials”
- [cosmetic] “ray_disagg_pd_deploy” reworded the description of “credentials”
- [cosmetic] “kserve_generate_yaml” reworded the description of “credentials”
- [cosmetic] “agent_agentic_serving_helm_values” reworded the description of “credentials”
- [cosmetic] “wandb_list_projects” reworded the description of “credentials”
- [cosmetic] “ray_submit_job” reworded the description of “credentials”
- [cosmetic] “phoenix_test” reworded the description of “credentials”
- [cosmetic] “ollama_generate” reworded the description of “credentials”
- [cosmetic] “ollama_chat” reworded the description of “credentials”
- [cosmetic] “vllm_sleep” reworded the description of “credentials”
- [cosmetic] “create_postgresql_connection” reworded the description of “credentials”
- [cosmetic] “agent_plan” reworded the description of “credentials”
- [cosmetic] “preflight_network_check” reworded the description of “credentials”
- [cosmetic] “langfuse_export_training_data” reworded the description of “credentials”
- [cosmetic] “ml_vllm_lora_unload” reworded the description of “credentials”
- [cosmetic] “langfuse_trace” reworded the description of “credentials”
- [cosmetic] “qdrant_test” reworded the description of “credentials”
- [cosmetic] “ml_vllm_lora_load” reworded the description of “credentials”
- [cosmetic] “governance_compliance_report” reworded the description of “credentials”
- [cosmetic] “k8s_info” reworded the description of “credentials”
- [cosmetic] “langfuse_test” reworded the description of “credentials”
- [cosmetic] “wandb_dashboard_status” reworded the description of “credentials”
- [cosmetic] “ray_wide_ep_deploy” reworded the description of “credentials”
- [cosmetic] “deepeval_evaluate” reworded the description of “credentials”
- [cosmetic] “up” reworded the description of “credentials”
- [cosmetic] “ollama_list” reworded the description of “credentials”
- [cosmetic] “qdrant_create_collection” reworded the description of “credentials”
- [cosmetic] “orchestrator_infer” reworded the description of “credentials”
- [cosmetic] “hf_hardware_recommend” reworded the description of “credentials”
- [cosmetic] “upsert_database” reworded the description of “credentials”
- [cosmetic] “agent_mem0_configure” reworded the description of “credentials”
- [cosmetic] “orchestrator_evict” reworded the description of “credentials”
- [cosmetic] “gpu_topology” reworded the description of “credentials”
- [cosmetic] “langgraph_evaluation_workflow” reworded the description of “credentials”
- [cosmetic] “qdrant_count” reworded the description of “credentials”
- [cosmetic] “wandb_setup_alerts” reworded the description of “credentials”
- [cosmetic] “inferx_configure” reworded the description of “credentials”
- [cosmetic] “agent_mem0_test” reworded the description of “credentials”

### 2026-09-09 (score 58, +1)

We did not load change detail this far back for this component, so this day may have recorded more than is shown.

### 2026-09-08 (score 57)

First indexed and scored.

## MCP tools (245)

### `provision_gpu` (~98 tokens)

Provision GPU instances for optimal parallel efficiency

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…
- `gpu_type` (string, required): GPU type to provision

### `preflight_report` (~113 tokens)

Generate full preflight validation report with pass/warn/fail per check. Covers GPU drivers, CUDA, NCCL, RDMA, network, disk, and Docker.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…

### `preflight_gpu_check` (~120 tokens)

GPU-specific preflight validation: NVIDIA drivers, CUDA version, GPU count, NCCL, NVLink topology, NCU stall-signature profiling, and adversarial config verification (V1-V3).

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…

### `preflight_network_check` (~105 tokens)

Network-specific preflight validation: RDMA availability, InfiniBand status, inter-node bandwidth, latency matrix, firewall rules.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…

### `local_scan` (~98 tokens)

Scan local machine and network for available GPU devices. Returns total VRAM pool for local-first provisioning.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…

### `status` (~86 tokens)

View all instances and costs with state optimization

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…

### `manage_instance` (~107 tokens)

Manage GPU instances (stop/start/terminate)

Input parameters:

- `action` (string, required): Action to perform
- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…
- `instance_id` (string, required): Instance ID

### `analytics` (~81 tokens)

Get cost analytics

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…

### `optimize` (~84 tokens)

Find cheaper alternatives for running instances

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…

### `setup_provider` (~94 tokens)

Get setup instructions for a provider

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…
- `provider` (string, required): Provider name

### `configure_provider` (~92 tokens)

Configure provider credentials

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…
- `provider` (string, required): Provider to configure

### `train` (~136 tokens)

Launch distributed training on provisioned GPU nodes. Supports torchrun, deepspeed, accelerate, and megatron. Use from_provision='latest' to auto-resolve node IPs from your last provision command.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…
- `script` (string, required): Training script path (e.g., train.py)

### `preflight` (~96 tokens)

Pre-training validation: GPU availability, NCCL, RDMA, drivers across all nodes.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…

### `stage` (~143 tokens)

Compress, chunk, checksum, and position datasets near compute. Supports local paths, S3/GCS URIs, HTTP URLs, and HuggingFace dataset names. Returns staging plan with agent recommendations.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…
- `dataset` (string, required): Dataset path, S3 URI, GCS URI, HTTP URL, or HuggingFace name

### `up` (~114 tokens)

CLI-native provisioning with manifest cache and drift detection. Use --fix-drift to detect and auto-fix drifted infrastructure.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…
- `job` (string, required): Job name for manifest tracking

### `rollback` (~118 tokens)

Explicit versioned rollback. Format: job@version (e.g., llama3@v3).

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…
- `job_version` (string, required): Job@version string (e.g., llama3@v3)

### `manifests` (~85 tokens)

List cached manifests and versions for jobs.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…

### `run_workflow` (~120 tokens)

Run a declarative YAML workflow that chains multiple Terradev commands (provision → preflight → train → monitor → checkpoint). Returns step-by-step execution status with cost estimates and confirmation gates for expensive operations.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…

### `active_context` (~109 tokens)

Get current Terradev state: running training jobs, active instances, spend-to-date, alerts. Call this on session start to resume context from previous sessions.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…

### `governance_request_consent` (~145 tokens)

Request user consent for data movement across cloud regions. GDPR/SOC2 compliant consent tracking with audit trail.

Input parameters:

- `consent_type` (string, required): Consent type
- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…
- `dataset_name` (string, required): Dataset being moved
- `purpose` (string, required): Purpose of data movement
- `user_id` (string, required): User ID requesting consent

### `governance_record_consent` (~128 tokens)

Record a consent response (granted or denied) for a pending consent request.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…
- `granted` (boolean, required): Whether consent was granted
- `request_id` (string, required): Consent request ID
- `user_id` (string, required): User ID

### `governance_evaluate_opa` (~137 tokens)

Evaluate OPA (Open Policy Agent) policies for data access. Checks region restrictions, classification rules, and compliance requirements.

Input parameters:

- `action` (string, required): Action to evaluate
- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…
- `dataset_name` (string, required): Dataset name
- `user_id` (string, required): User ID to evaluate

### `governance_move_data` (~158 tokens)

Move data with full governance audit trail. Requires prior consent and OPA policy approval. Tracks integrity, encryption, and compliance.

Input parameters:

- `consent_request_id` (string, required): Approved consent request ID
- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…
- `dataset_name` (string, required): Dataset to move
- `source_location` (string, required): Source location
- `target_location` (string, required): Target location
- `user_id` (string, required): User ID

### `governance_movement_history` (~96 tokens)

Get data movement audit log. Filter by user, dataset, or time range.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…

### `governance_compliance_report` (~154 tokens)

Generate comprehensive compliance report: consent stats, policy evaluations, data movements, violations. For GDPR/SOC2/HIPAA audits.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…
- `end_date` (string, required): End date (ISO format, e.g. 2025-12-31)
- `start_date` (string, required): Start date (ISO format, e.g. 2025-01-01)

### `inferx_deploy` (~127 tokens)

Deploy model to InferX serverless platform

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…
- `gpu_type` (string, required): GPU type for inference
- `model` (string, required): Model ID (e.g., meta-llama/Llama-2-7b-hf)

### `inferx_status` (~85 tokens)

Check InferX endpoint status

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…

### `inferx_list` (~85 tokens)

List deployed InferX models

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…

### `inferx_optimize` (~87 tokens)

Get cost analysis for inference endpoints

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…

### `inferx_configure` (~100 tokens)

Configure InferX serverless platform credentials.

Input parameters:

- `api_key` (string, required): InferX API key
- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…

### `inferx_delete` (~99 tokens)

Delete an InferX model deployment.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…
- `model_id` (string, required): Model deployment ID to delete

### `inferx_usage` (~95 tokens)

Get InferX account usage statistics: requests, cost, GPU hours, latency.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…

### `inferx_quote` (~89 tokens)

Get InferX pricing quotes for a GPU type.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…

### `infer_route` (~129 tokens)

Semantic-aware inference routing. Analyzes query content across 6 signal dimensions (modality, complexity, domain, language, safety, keywords), applies NUMA-aware endpoint scoring, and selects the optimal inference endpoint. Uses DAG parallel execution for signal extraction.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…

### `infer_route_disagg` (~171 tokens)

Disaggregated Prefill/Decode routing (DistServe architecture). Splits LLM inference into compute-bound prefill phase (routed to FLOPS-optimized GPUs like H100 SXM) and memory-bound decode phase (routed to bandwidth-optimized GPUs like MI300X). Tracks KV cache handoffs between endpoint pairs.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…
- `model` (string, required): Model name for disaggregated pair selection (e.g., llama-3-70b)

### `infer_failover` (~111 tokens)

Run health checks and auto-failover for inference endpoints. If a primary endpoint is unhealthy and has a backup configured, traffic automatically shifts to the backup provider.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…

### `gpu_topology` (~185 tokens)

GPU NUMA topology report with intra-GPU XCD (Accelerated Compute Die) awareness. Models MI300X (8 XCDs, 192GB HBM3), MI300A (6 XCDs, 128GB), H200 (unified 141GB HBM3e), H100 (80GB). Reports PCIe locality (PIX/PXB/PHB/SYS), GPU-NIC pairing, SR-IOV VF status, and generates XCD-aware NCCL/AITER environment variables.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…

### `vllm_start` (~136 tokens)

Start a vLLM inference server on a remote instance via SSH/systemd. Supports Multi-LoRA, Sleep Mode, KV Offloading, Speculative Decoding.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…
- `instance_ip` (string, required): Target instance IP
- `model` (string, required): HuggingFace model ID

### `vllm_stop` (~102 tokens)

Stop a vLLM server on a remote instance.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…
- `instance_ip` (string, required): Target instance IP

### `vllm_inference` (~136 tokens)

Test inference against a running vLLM endpoint (completions or chat).

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…
- `endpoint` (string, required): vLLM endpoint URL (e.g. http://10.0.0.1:8000)
- `model` (string, required): Model name to use

### `vllm_info` (~124 tokens)

Get vLLM server info: loaded models, config, and health status.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…
- `endpoint` (string, required): vLLM endpoint URL (e.g. http://10.0.0.1:8000)

### `vllm_sleep` (~123 tokens)

Put a vLLM server to sleep. Level 1: offload to CPU (fast wake). Level 2: discard weights (minimal RAM).

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…
- `endpoint` (string, required): vLLM endpoint URL

### `vllm_wake` (~116 tokens)

Wake a sleeping vLLM server. For Level 2 sleep, also reloads weights and resets prefix cache.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…
- `endpoint` (string, required): vLLM endpoint URL

### `sglang` (~136 tokens)

Complete SGLang optimization stack with workload-specific auto-tuning for 7 workload types: agentic chat, batch inference, low latency, MoE models, PD disaggregated, structured output, and RAG.

Input parameters:

- `action` (string, required): SGLang action to perform
- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…

### `sglang_start` (~129 tokens)

[DEPRECATED] Use sglang action='start' instead. Start an SGLang inference server on a remote instance.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…
- `instance_ip` (string, required): Target instance IP
- `model` (string, required): HuggingFace model ID

### `sglang_stop` (~116 tokens)

[DEPRECATED] Use sglang action='stop' instead. Stop an SGLang server on a remote instance.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…
- `instance_ip` (string, required): Target instance IP

### `sglang_inference` (~145 tokens)

[DEPRECATED] Use sglang action='inference' instead. Test inference against a running SGLang endpoint.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…
- `endpoint` (string, required): SGLang endpoint URL (e.g. http://10.0.0.1:8000)
- `model` (string, required): Model name to use

### `ollama_list` (~88 tokens)

List models available on an Ollama server.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…

### `ollama_pull` (~129 tokens)

Pull a model to an Ollama server on a remote instance.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…
- `instance_ip` (string, required): Target instance IP
- `model` (string, required): Model name (e.g. llama3.2, deepseek-r1, codellama)

### `ollama_generate` (~110 tokens)

Generate text using an Ollama model (non-chat completions).

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…
- `model` (string, required): Model name
- `prompt` (string, required): Prompt text

### `ollama_chat` (~115 tokens)

Chat with an Ollama model using the chat/completions API.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…
- `messages` (array, required): Chat messages [{role, content}]
- `model` (string, required): Model name

### `ollama_model_info` (~104 tokens)

Get detailed information about an Ollama model (parameters, template, license).

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…
- `model` (string, required): Model name

### `vllm_auto_optimize` (~122 tokens)

Automatically optimize vLLM configuration by analyzing workload patterns. Selects optimal settings for the 6 critical knobs based on live endpoint metrics or sample request files.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…
- `model` (string, required): Model name

### `vllm_analyze` (~123 tokens)

Analyze a running vLLM server's workload and return specific optimization recommendations with before/after comparisons.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…
- `endpoint` (string, required): vLLM endpoint URL (e.g. http://localhost:8000)

### `vllm_benchmark` (~116 tokens)

Benchmark a vLLM endpoint with concurrent requests. Returns throughput (req/s), success rate, and total latency.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…
- `endpoint` (string, required): vLLM endpoint URL

### `ollama_ps` (~86 tokens)

List currently running Ollama models.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…

### `langfuse_configure` (~132 tokens)

Configure Langfuse credentials (public key, secret key, host URL).

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…
- `public_key` (string, required): Langfuse public key (pk-lf-...)
- `secret_key` (string, required): Langfuse secret key (sk-lf-...)

### `langfuse_test` (~90 tokens)

Test Langfuse connectivity and list accessible projects.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…

### `langfuse_traces` (~91 tokens)

List recent LLM traces from Langfuse.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…

### `langfuse_trace` (~103 tokens)

Get a single Langfuse trace with all observations/spans.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…
- `trace_id` (string, required): Trace ID

### `langfuse_scores` (~96 tokens)

List evaluation scores from Langfuse, optionally filtered by trace or score name.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…

### `langfuse_score` (~138 tokens)

Create an evaluation score for a Langfuse trace (e.g. quality, accuracy, relevance).

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…
- `name` (string, required): Score name (e.g. quality, accuracy)
- `trace_id` (string, required): Trace ID to score
- `value` (number, required): Numeric score value

### `langfuse_datasets` (~93 tokens)

List Langfuse datasets for evaluation and fine-tuning.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…

### `langfuse_export_training_data` (~104 tokens)

Export Langfuse traces as instruction/response pairs for LoRA fine-tuning. Filters by quality score.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…

### `langfuse_quality` (~93 tokens)

Get aggregated quality metrics from Langfuse scores for drift detection.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…

### `langfuse_otel_env` (~101 tokens)

Print OTEL environment variables for instrumenting LLM apps to send traces to Langfuse.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…

### `langfuse_k8s` (~94 tokens)

Generate Kubernetes deployment manifest for self-hosted Langfuse.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…

### `k8s_create` (~116 tokens)

Create Kubernetes cluster with GPU nodes for optimal multi-cloud deployment

Input parameters:

- `cluster_name` (string, required): Name of the cluster
- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…
- `gpu_type` (string, required): GPU type for nodes

### `k8s_list` (~84 tokens)

List Kubernetes clusters

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…

### `k8s_info` (~99 tokens)

Get information about a specific cluster

Input parameters:

- `cluster_name` (string, required): Name of the cluster
- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…

### `k8s_destroy` (~99 tokens)

Destroy a Kubernetes cluster

Input parameters:

- `cluster_name` (string, required): Name of the cluster to destroy
- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…

### `helm_generate` (~104 tokens)

Generate Helm charts from workload specifications.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…
- `image` (string, required): Container image
- `workload` (string, required): Workload type

### `kserve_generate_yaml` (~170 tokens)

Generate a GPU-aware KServe InferenceService YAML manifest with NUMA pinning, resource limits derived from model size and VRAM, and topology hints.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…
- `gpu_type` (string, required): GPU type (e.g. A100, H100)
- `model_name` (string, required): Model name for the InferenceService
- `model_uri` (string, required): Model storage URI (e.g. s3://bucket/model or gs://bucket/model)

### `kserve_list` (~90 tokens)

List KServe InferenceServices in a Kubernetes namespace.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…

### `kserve_status` (~109 tokens)

Get detailed status of a KServe InferenceService including readiness, traffic split, and URL.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…
- `name` (string, required): InferenceService name

### `k8s_gpu_operator_install` (~119 tokens)

Install NVIDIA GPU Operator on a Kubernetes cluster. Configures driver containers, device plugin, DCGM exporter, and GPU Feature Discovery.

Input parameters:

- `cluster_name` (string, required): Target cluster name
- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…

### `k8s_device_plugin` (~110 tokens)

Configure Kubernetes GPU device plugin settings: time-slicing, MIG strategy, and resource naming.

Input parameters:

- `cluster_name` (string, required): Target cluster name
- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…

### `k8s_mig_configure` (~136 tokens)

Configure Multi-Instance GPU (MIG) partitioning on A100/H100 GPUs. Splits a single GPU into isolated instances for multi-tenant workloads.

Input parameters:

- `cluster_name` (string, required): Target cluster name
- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…
- `mig_profile` (string, required): MIG profile

### `k8s_time_slicing` (~115 tokens)

Configure GPU time-slicing for Kubernetes. Allows multiple pods to share a single GPU with configurable oversubscription.

Input parameters:

- `cluster_name` (string, required): Target cluster name
- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…

### `hf_space_deploy` (~122 tokens)

Deploy model to HuggingFace Spaces

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…
- `model_id` (string, required): HuggingFace model ID
- `space_name` (string, required): Name of the Space
- `template` (string, required): Template type

### `hf_space_status` (~99 tokens)

Get HuggingFace Space deployment status.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…
- `space_name` (string, required): Space name to check

### `ray_status` (~94 tokens)

Get Ray cluster status including node count, resources, memory, and running jobs.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…

### `ray_start` (~95 tokens)

Start a Ray cluster (head node or worker). For distributed ML training and inference.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…

### `ray_stop` (~87 tokens)

Stop the Ray cluster on the current node.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…

### `ray_submit_job` (~105 tokens)

Submit a job script to the Ray cluster for distributed execution.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…
- `script` (string, required): Path to the Python script to submit

### `ray_list_jobs` (~87 tokens)

List all running Ray jobs and tasks.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…

### `ray_wide_ep_deploy` (~154 tokens)

Generate a Ray Serve LLM Wide-EP (Expert Parallel) deployment for MoE models. Returns Python script and config for distributed MoE serving with EPLB and DeepEP.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…
- `model_id` (string, required): HuggingFace model ID (e.g. zai-org/GLM-5-FP8, deepseek-ai/DeepSeek-V3)

### `ray_disagg_pd_deploy` (~132 tokens)

Generate a Ray Serve LLM disaggregated Prefill/Decode deployment. Splits inference into compute-bound prefill and memory-bound decode phases with KV cache transfer via NIXL.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…
- `model_id` (string, required): HuggingFace model ID

### `ray_parallelism_strategy` (~133 tokens)

Compute optimal TP/DP/EP parallelism strategy for a given MoE model and GPU count. Returns recommended configuration with rationale.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…
- `model_id` (string, required): HuggingFace model ID (e.g. zai-org/GLM-5-FP8)

### `wandb_list_projects` (~105 tokens)

List all Weights & Biases projects for the configured entity.

Input parameters:

- `api_key` (string, required): W&B API key
- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…

### `wandb_list_runs` (~119 tokens)

List runs in a W&B project with status, metrics summary, and config.

Input parameters:

- `api_key` (string, required): W&B API key
- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…
- `project` (string, required): W&B project name

### `wandb_run_details` (~119 tokens)

Get detailed info, metrics, and artifacts for a specific W&B run.

Input parameters:

- `api_key` (string, required): W&B API key
- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…
- `run_id` (string, required): W&B run ID

### `mlflow_list_experiments` (~105 tokens)

List MLflow experiments on the configured tracking server.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…
- `tracking_uri` (string, required): MLflow tracking server URI

### `mlflow_log_run` (~142 tokens)

Log a Terradev training run to MLflow with auto-injected GPU type, provider, cost/hr, and duration as params.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…
- `experiment_name` (string, required): MLflow experiment name
- `run_name` (string, required): Run display name
- `tracking_uri` (string, required): MLflow tracking server URI

### `mlflow_register_model` (~136 tokens)

Register a trained model in the MLflow model registry with Terradev provenance tags.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…
- `model_name` (string, required): Model registry name
- `run_id` (string, required): MLflow run ID that produced the model
- `tracking_uri` (string, required): MLflow tracking server URI

### `dvc_status` (~111 tokens)

Get DVC repository status: tracked files, remotes, and changes since last commit.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…
- `repo_path` (string, required): Path to the DVC repository

### `dvc_diff` (~115 tokens)

Show DVC diff between two revisions (e.g. training checkpoints). Shows added, modified, deleted files.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…
- `repo_path` (string, required): Path to the DVC repository

### `dvc_stage_checkpoint` (~137 tokens)

Atomic checkpoint staging: DVC add + push + git commit in one operation. Promotes a training checkpoint to versioned storage.

Input parameters:

- `checkpoint_path` (string, required): Path to the checkpoint file/directory to stage
- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…
- `repo_path` (string, required): Path to the DVC repository

### `dvc_push` (~105 tokens)

Push DVC-tracked data to the configured remote storage.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…
- `repo_path` (string, required): Path to the DVC repository

### `hf_list_models` (~120 tokens)

Search and browse HuggingFace Hub models. Filter by author, task, library. Returns model ID, downloads, likes, and tags.

Input parameters:

- `api_key` (string, required): HuggingFace API token
- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…

### `hf_list_datasets` (~109 tokens)

Search and browse HuggingFace Hub datasets. Filter by author and search query.

Input parameters:

- `api_key` (string, required): HuggingFace API token
- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…

### `hf_model_info` (~146 tokens)

Get detailed model info: architecture, size, downloads, license, tags, pipeline_tag, and model card.

Input parameters:

- `api_key` (string, required): HuggingFace API token
- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…
- `model_id` (string, required): HuggingFace model ID (e.g. meta-llama/Llama-3.3-70B-Instruct)

### `hf_create_endpoint` (~161 tokens)

Create a HuggingFace Inference Endpoint (paid GPU endpoint). Supports custom GPU types, regions, and scaling.

Input parameters:

- `api_key` (string, required): HuggingFace API token
- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…
- `endpoint_name` (string, required): Endpoint name
- `instance_type` (string, required): Instance type (e.g. nvidia-a100, nvidia-l4)
- `model_id` (string, required): HuggingFace model ID

### `hf_list_endpoints` (~111 tokens)

List all active HuggingFace Inference Endpoints with status, URL, and cost.

Input parameters:

- `api_key` (string, required): HuggingFace API token
- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…

### `hf_endpoint_info` (~124 tokens)

Get detailed info about a specific HuggingFace Inference Endpoint: status, URL, scaling config, cost.

Input parameters:

- `api_key` (string, required): HuggingFace API token
- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…
- `endpoint_name` (string, required): Endpoint name

### `hf_delete_endpoint` (~113 tokens)

Delete a HuggingFace Inference Endpoint.

Input parameters:

- `api_key` (string, required): HuggingFace API token
- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…
- `endpoint_name` (string, required): Endpoint name to delete

### `hf_endpoint_infer` (~135 tokens)

Run inference on a HuggingFace Inference Endpoint. Supports text generation, embeddings, and custom inputs.

Input parameters:

- `api_key` (string, required): HuggingFace API token
- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…
- `endpoint_name` (string, required): Endpoint name
- `inputs` (string, required): Input text or prompt

### `hf_smart_template` (~149 tokens)

Auto-generate an optimized deployment template for any HuggingFace model. Analyzes model size, architecture, and quantization to select optimal hardware and generate ready-to-deploy configs.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…
- `model_id` (string, required): HuggingFace model ID (e.g. meta-llama/Llama-3.3-70B-Instruct)

### `hf_hardware_recommend` (~119 tokens)

Get hardware recommendation with cost breakdown for any HuggingFace model. Returns optimal GPU type, estimated cost, and performance score.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…
- `model_id` (string, required): HuggingFace model ID

### `hf_hardware_compare` (~116 tokens)

Compare all hardware options for a HuggingFace model. Returns side-by-side cost, performance, and compatibility analysis.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…
- `model_id` (string, required): HuggingFace model ID

### `langchain_create_workflow` (~117 tokens)

Create a LangChain workflow.

Input parameters:

- `api_key` (string, required): LangChain API key
- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…
- `workflow_config` (object, required): Workflow configuration (name, steps, model, tools)

### `langchain_create_sglang_pipeline` (~138 tokens)

Create an SGLang model-serving pipeline via LangChain. Connects LangChain agents to SGLang inference endpoints.

Input parameters:

- `api_key` (string, required): LangChain API key
- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…
- `pipeline_config` (object, required): Pipeline config (model, endpoint, temperature, max_tokens)

### `langgraph_create_workflow` (~131 tokens)

Create a LangGraph stateful workflow with monitoring. Supports agent graphs, tool calling, and state persistence.

Input parameters:

- `api_key` (string, required): LangChain API key
- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…
- `graph_config` (object, required): Graph configuration (nodes, edges, state_schema)

### `langgraph_orchestrator_worker` (~138 tokens)

Create an orchestrator-worker pattern workflow in LangGraph. The orchestrator delegates tasks to specialized worker agents.

Input parameters:

- `api_key` (string, required): LangChain API key
- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…
- `workflow_config` (object, required): Orchestrator-worker config (orchestrator_prompt, workers, routing_strategy)

### `langgraph_evaluation_workflow` (~135 tokens)

Create an evaluator-optimizer workflow in LangGraph. Generates outputs, evaluates quality, and iteratively improves.

Input parameters:

- `api_key` (string, required): LangChain API key
- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…
- `evaluation_config` (object, required): Evaluation config (generator_prompt, evaluator_criteria, max_iterations)

### `langgraph_workflow_status` (~117 tokens)

Get the status and metrics of a LangGraph workflow execution.

Input parameters:

- `api_key` (string, required): LangChain API key
- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…
- `workflow_id` (string, required): Workflow ID to check

### `wandb_create_dashboard` (~125 tokens)

Create a custom W&B dashboard with GPU metrics, training loss, and cost panels.

Input parameters:

- `api_key` (string, required): W&B API key
- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…
- `dashboard_config` (object, required): Dashboard config (name, panels, metrics)

### `wandb_create_terradev_dashboard` (~119 tokens)

Auto-create a Terradev-specific W&B dashboard with GPU utilization, cost tracking, training metrics, and infrastructure panels.

Input parameters:

- `api_key` (string, required): W&B API key
- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…

### `wandb_create_report` (~125 tokens)

Create a W&B report with custom sections, charts, and narrative text.

Input parameters:

- `api_key` (string, required): W&B API key
- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…
- `report_config` (object, required): Report config (title, sections, metrics, description)

### `wandb_create_terradev_report` (~117 tokens)

Auto-generate a Terradev infrastructure report: GPU costs, provider comparison, training efficiency, and recommendations.

Input parameters:

- `api_key` (string, required): W&B API key
- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…

### `wandb_setup_alerts` (~131 tokens)

Set up custom W&B alerts for GPU metrics: cost thresholds, utilization drops, training anomalies.

Input parameters:

- `alert_config` (object, required): Alert config (metric, threshold, condition, notification_channel)
- `api_key` (string, required): W&B API key
- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…

### `wandb_create_terradev_alerts` (~123 tokens)

Auto-create standard Terradev alerts: GPU cost > budget, utilization < 50%, training loss spike, straggler detection.

Input parameters:

- `api_key` (string, required): W&B API key
- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…

### `wandb_dashboard_status` (~108 tokens)

Get comprehensive W&B monitoring overview: dashboards, reports, alerts, active runs.

Input parameters:

- `api_key` (string, required): W&B API key
- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…

### `phoenix_test` (~93 tokens)

Test connection to Arize Phoenix server. Returns collector endpoint and project count.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…

### `phoenix_projects` (~85 tokens)

List Phoenix projects (trace namespaces).

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…

### `phoenix_spans` (~112 tokens)

List recent spans for a Phoenix project. Supports SpanQuery DSL filters like "span_kind == 'RETRIEVER'" or "status_code == 'ERROR'".

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…

### `phoenix_trace` (~111 tokens)

View full execution tree for a specific trace ID. Shows span hierarchy, latencies, and token counts.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…
- `trace_id` (string, required): Trace ID to inspect

### `phoenix_otel_env` (~94 tokens)

Generate OpenTelemetry environment variables for instrumenting serving pods with Phoenix tracing.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…

### `phoenix_snippet` (~92 tokens)

Generate Python instrumentation snippet for adding Phoenix tracing to LLM applications.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…

### `phoenix_k8s` (~93 tokens)

Generate Kubernetes deployment manifest for self-hosted Arize Phoenix server.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…

### `guardrails_test` (~88 tokens)

Test connection to NeMo Guardrails server.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…

### `guardrails_chat` (~121 tokens)

Send a message through NeMo Guardrails and return the safety-filtered response. Applies topical, jailbreak, PII, and factcheck rails.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…
- `message` (string, required): Message to send through guardrails

### `guardrails_generate_config` (~104 tokens)

Generate default Colang 2.x guardrails configuration files (topical, jailbreak, PII, factcheck rails).

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…

### `guardrails_k8s` (~99 tokens)

Generate Kubernetes deployment manifest for NeMo Guardrails server (standalone or sidecar mode).

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…

### `qdrant_test` (~96 tokens)

Test connection to Qdrant vector database. Returns cluster info and collection count.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…

### `qdrant_collections` (~95 tokens)

List all Qdrant vector collections with their point counts and configurations.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…

### `qdrant_create_collection` (~99 tokens)

Create a Qdrant vector collection. Auto-configures vector dimensions from embedding model name.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…

### `qdrant_info` (~92 tokens)

Get detailed info and stats for a Qdrant collection.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…

### `qdrant_count` (~92 tokens)

Count points (vectors) in a Qdrant collection.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…

### `qdrant_k8s` (~97 tokens)

Generate Kubernetes StatefulSet manifest for self-hosted Qdrant vector database.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…

### `deepeval_run` (~92 tokens)

Run a DeepEval test suite from a Python test file.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…

### `deepeval_metrics` (~90 tokens)

List available DeepEval metrics for LLM evaluation.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…

### `deepeval_evaluate` (~141 tokens)

Evaluate a single LLM output with a DeepEval metric (AnswerRelevancy, Faithfulness, Hallucination, etc.).

Input parameters:

- `actual_output` (string, required): Actual LLM output
- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…
- `input` (string, required): Test input/prompt
- `metric` (string, required): DeepEval metric name

### `deepeval_init` (~88 tokens)

Generate a starter DeepEval test file.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…

### `egress_cheapest_route` (~177 tokens)

Find the cheapest egress route between cloud providers/regions for model weights or dataset transfer. Supports multi-hop routing.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…
- `dest_provider` (string, required): Destination cloud provider
- `dest_region` (string, required): Destination region
- `size_gb` (number, required): Transfer size in GB
- `source_provider` (string, required): Source cloud provider (e.g. aws, gcp, azure)
- `source_region` (string, required): Source region (e.g. us-east-1)

### `egress_optimize_staging` (~163 tokens)

Optimize dataset or model staging across regions by finding the cheapest transfer plan. Integrates with the dataset stager for parallel uploads.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…
- `size_gb` (number, required): Approximate data size in GB
- `source_uri` (string, required): Source data URI (s3://, gs://, local path, or HF dataset ID)
- `target_regions` (array, required): Target regions as provider:region strings

### `orchestrator_start` (~93 tokens)

Start the model orchestrator for multi-model GPU sharing with eviction policies.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…

### `orchestrator_register` (~109 tokens)

Register a model with the orchestrator.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…
- `model_id` (string, required): Model identifier
- `model_path` (string, required): Path to model weights

### `orchestrator_load` (~97 tokens)

Load a model into GPU memory.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…
- `model_id` (string, required): Model to load

### `orchestrator_evict` (~100 tokens)

Evict a model from GPU memory.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…
- `model_id` (string, required): Model to evict

### `orchestrator_status` (~90 tokens)

Get orchestrator and model status including GPU memory utilization.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…

### `orchestrator_infer` (~103 tokens)

Test inference with a model via the orchestrator.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…
- `model_id` (string, required): Model to run inference on

### `warm_pool_start` (~112 tokens)

Start the warm pool manager for intelligent model pre-warming. 5 strategies: traffic_based, time_based, priority_based, cost_optimized, latency_optimized.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…

### `warm_pool_status` (~96 tokens)

Get warm pool status: hit rate, cold starts, memory saved, cost saved.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…

### `price_intel` (~138 tokens)

GPU price intelligence with quantitative analytics. Computes delta (rate of change), gamma (acceleration), and annualized realized volatility on GPU spot/on-demand prices across 21+ providers. Identifies cheapest time windows and provider arbitrage opportunities.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…
- `gpu_type` (string, required): GPU type to analyze

### `cost_analyze` (~101 tokens)

Deep cost analysis of current GPU infrastructure: per-provider breakdown, utilization efficiency, waste identification, and optimization potential.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…

### `cost_optimize_recommend` (~105 tokens)

Generate actionable cost optimization recommendations: spot migration, GPU right-sizing, provider arbitrage, idle shutdown, and density packing.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…

### `cost_simulate` (~115 tokens)

Simulate cost optimization scenarios with ROI projections. Compare current vs optimized infrastructure costs.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…
- `scenario` (object, required): Scenario config (gpu_type, provider, count, spot, hours)

### `price_trends` (~117 tokens)

Get GPU price trend analysis with delta (rate of change), gamma (acceleration), and annualized volatility. Identifies cheapest time windows.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…
- `gpu_type` (string, required): GPU type

### `price_spot_risk` (~111 tokens)

Spot instance risk assessment per provider. Returns interruption probability, mean time to interruption, and recommended mitigation.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…
- `gpu_type` (string, required): GPU type

### `moe_deploy` (~257 tokens)

Deploy Mixture-of-Experts models with production-ready cluster templates. Auto-applies vLLM cost optimizations (KV cache offloading for up to 9x throughput, MTP speculative decoding for up to 2.8x speed, sleep mode for 18-200x faster restarts). Supports GLM-5, Qwen 3.5, Mistral Large 3, DeepSeek V4, Llama 5. Configures NVLink topology, tensor parallelism, FP8 quantization, vLLM/SGLang backends, and GPU-aware HPA autoscaling.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…
- `gpu_type` (string, required): GPU type for MoE serving
- `model_id` (string, required): HuggingFace model ID (e.g., zai-org/GLM-5-FP8, Qwen/Qwen3.5-397B-A17B)

### `train_status` (~94 tokens)

List all training jobs and their state (created, running, completed, failed).

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…

### `train_monitor` (~110 tokens)

Real-time GPU monitoring for training jobs. Shows utilization, memory, temperature, power, and cost.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…
- `job_id` (string, required): Job ID to monitor

### `checkpoint_list` (~96 tokens)

List all checkpoints for a training job.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…
- `job_id` (string, required): Job ID

### `checkpoint_save` (~100 tokens)

Manually trigger a checkpoint save for a running training job.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…
- `job_id` (string, required): Job ID

### `train_stop` (~106 tokens)

Stop a running training job. Kills training processes on all nodes in parallel.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…
- `job_id` (string, required): Job ID to stop

### `train_resume` (~108 tokens)

Resume a training job from its latest checkpoint. Rebuilds config with topology revalidation.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…
- `job_id` (string, required): Job ID to resume

### `checkpoint_restore` (~97 tokens)

Restore a specific checkpoint for a training job.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…
- `job_id` (string, required): Job ID

### `checkpoint_promote` (~130 tokens)

Promote a checkpoint to a final model path for serving.

Input parameters:

- `checkpoint_id` (string, required): Checkpoint ID to promote
- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…
- `dest` (string, required): Destination path (e.g., /models/final)
- `job_id` (string, required): Job ID

### `checkpoint_delete` (~105 tokens)

Delete a checkpoint.

Input parameters:

- `checkpoint_id` (string, required): Checkpoint ID to delete
- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…
- `job_id` (string, required): Job ID

### `lora_list` (~131 tokens)

List LoRA adapters loaded on a running vLLM endpoint. Shows base models and hot-loaded fine-tuned adapters.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…
- `endpoint` (string, required): vLLM endpoint URL (e.g. http://10.0.0.1:8000)

### `lora_add` (~195 tokens)

Hot-load a LoRA adapter onto a running vLLM endpoint. The adapter becomes immediately available as a model name for inference requests. Uses vLLM's fused_moe_lora kernel for 454% higher output tokens/sec on MoE models.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…
- `endpoint` (string, required): vLLM endpoint URL (e.g. http://10.0.0.1:8000)
- `name` (string, required): Adapter name (becomes the model name in API requests)
- `path` (string, required): Path to adapter weights (local path or HuggingFace ID)

### `lora_remove` (~141 tokens)

Hot-unload a LoRA adapter from a running vLLM endpoint. Frees GPU memory for other adapters.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…
- `endpoint` (string, required): vLLM endpoint URL (e.g. http://10.0.0.1:8000)
- `name` (string, required): Adapter name to unload

### `training_config_generate` (~127 tokens)

Generate a complete training configuration from a declarative spec. Auto-detects framework, sets optimal parallelism, and configures distributed training.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…
- `name` (string, required): Training job name
- `script` (string, required): Training script path

### `training_launch_distributed` (~126 tokens)

Full distributed training launch with framework auto-detection, topology validation, and monitoring. Combines preflight + train + monitor in one operation.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…
- `name` (string, required): Job name
- `script` (string, required): Training script

### `train_snapshot` (~128 tokens)

Get complete training monitoring snapshot: GPU metrics (utilization, memory, temp, power), training metrics (loss, grad_norm, lr, throughput), straggler detection, and cost estimate.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…
- `job_id` (string, required): Training job ID

### `train_detect_stragglers` (~115 tokens)

Detect straggler nodes in distributed training. Identifies GPUs with significantly lower utilization that slow the whole job.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…
- `job_id` (string, required): Training job ID

### `agent_agentic_serving_configure` (~90 tokens)

Configure agentic inference serving settings.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…

### `agent_agentic_serving_helm_values` (~93 tokens)

Print Helm values for agentic inference deployment.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…

### `agent_agentic_serving_k8s` (~95 tokens)

Print K8s deployment manifests for agentic inference.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…

### `agent_agentic_serving_launch_args` (~92 tokens)

Print engine launch arguments for copy-paste.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…

### `agent_agentic_serving_lmcache_env` (~91 tokens)

Print LMCache environment variables.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…

### `agent_agentic_serving_show_config` (~90 tokens)

Show current agentic serving configuration.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…

### `agent_cost` (~99 tokens)

Show real-time cost breakdown for a fleet by tier.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…
- `fleet_id` (string, required): Fleet ID

### `agent_deploy` (~89 tokens)

Provision a heterogeneous agent fleet across all tiers simultaneously.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…

### `agent_langchain_create_langgraph` (~98 tokens)

Create a LangGraph workflow.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…
- `graph_name` (string, required): graph_name

### `agent_langchain_create_pipeline` (~98 tokens)

Create an SGLang pipeline.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…
- `pipeline_name` (string, required): pipeline_name

### `agent_langchain_create_workflow` (~98 tokens)

Create a LangChain workflow.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…
- `workflow_name` (string, required): workflow_name

### `agent_langchain_test` (~87 tokens)

Test connection to LangChain service.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…

### `agent_langgraph_create_workflow` (~107 tokens)

Create a LangGraph workflow.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…
- `type` (string, required): Workflow type
- `workflow_name` (string, required): workflow_name

### `agent_langgraph_deploy` (~95 tokens)

Deploy a workflow.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…
- `workflow_name` (string, required): workflow_name

### `agent_langgraph_status` (~94 tokens)

Get workflow status.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…
- `workflow_id` (string, required): workflow_id

### `agent_langgraph_test` (~87 tokens)

Test connection to LangGraph service.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…

### `agent_letta_chat` (~109 tokens)

Send a message to a Letta agent.

Input parameters:

- `agent_id` (string, required): Agent ID
- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…
- `message` (string, required): Message to send

### `agent_letta_create` (~98 tokens)

Create a new stateful Letta agent.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…
- `name` (string, required): Agent name

### `agent_letta_delete` (~96 tokens)

Delete a Letta agent.

Input parameters:

- `agent_id` (string, required): Agent ID
- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…

### `agent_letta_list` (~85 tokens)

List Letta agents.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…

### `agent_letta_remember` (~110 tokens)

Teach a Letta agent a durable fact.

Input parameters:

- `agent_id` (string, required): Agent ID
- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…
- `text` (string, required): Fact to remember

### `agent_letta_status` (~99 tokens)

Show the state of a Letta agent.

Input parameters:

- `agent_id` (string, required): Agent ID
- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…

### `agent_mem0_add` (~112 tokens)

Store a memory in Mem0 for an agent or user.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…
- `text` (string, required): Text to remember
- `user_id` (string, required): User ID

### `agent_mem0_configure` (~100 tokens)

Configure Mem0 credentials and defaults.

Input parameters:

- `api_key` (string, required): Mem0 API key
- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…

### `agent_mem0_delete` (~96 tokens)

Delete a memory by ID.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…
- `memory_id` (string, required): Memory ID

### `agent_mem0_forget` (~99 tokens)

Delete all memories matching an entity scope.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…
- `user_id` (string, required): User ID

### `agent_mem0_get` (~97 tokens)

Get a single memory by ID.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…
- `memory_id` (string, required): Memory ID

### `agent_mem0_list` (~97 tokens)

List memories for an entity scope.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…
- `user_id` (string, required): User ID

### `agent_mem0_search` (~104 tokens)

Search agent/user memories.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…
- `query` (string, required): Search query
- `user_id` (string, required): User ID

### `agent_mem0_test` (~86 tokens)

Test connection to Mem0.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…

### `agent_mem0_update` (~96 tokens)

Update a memory by ID.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…
- `memory_id` (string, required): Memory ID

### `agent_list` (~84 tokens)

List all known agent fleets.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…

### `agent_plan` (~101 tokens)

Plan a heterogeneous agent fleet without provisioning.

Input parameters:

- `agents` (integer, required): Number of concurrent agent loops to provision for
- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…

### `agent_scale` (~124 tokens)

Scale a single fleet tier up or down without affecting other tiers.

Input parameters:

- `count` (integer, required): New instance count for this tier
- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…
- `fleet_id` (string, required): Fleet ID
- `tier` (string, required): Tier to scale

### `agent_skill_attach` (~117 tokens)

Attach a skill.md to a Letta agent as a durable memory block.

Input parameters:

- `agent_id` (string, required): Letta agent ID
- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…
- `skill` (string, required): Path to skill.md

### `agent_skill_init` (~97 tokens)

Create a skill.md template for an agent.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…
- `name` (string, required): Skill name

### `agent_status` (~115 tokens)

Show live status of a fleet — tier health, KV hit rate, queue depth, cost.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…
- `fleet_id` (string, required): Fleet ID returned by 'terradev agent deploy'

### `agent_teardown` (~101 tokens)

Terminate all fleet instances and remove fleet state.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…
- `fleet_id` (string, required): Fleet ID to destroy

### `agent_vector_db_down` (~92 tokens)

Teardown a vector database provisioned for an agent fleet.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…

### `agent_vector_db_up` (~89 tokens)

Provision a vector database for an agent fleet.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…

### `ml_vllm_lora_link` (~119 tokens)

Load the active registry version of an adapter onto a vLLM server.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…
- `endpoint` (string, required): vLLM endpoint
- `name` (string, required): Registered adapter name

### `ml_vllm_lora_list` (~107 tokens)

List LoRA adapters currently loaded on a vLLM server.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…
- `endpoint` (string, required): vLLM endpoint

### `ml_vllm_lora_load` (~129 tokens)

Hot-load a LoRA adapter onto a running vLLM server.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…
- `endpoint` (string, required): vLLM endpoint
- `name` (string, required): Adapter name
- `path` (string, required): Local path to adapter weights

### `ml_vllm_lora_sync` (~121 tokens)

Synchronize an adapter from the registry across multiple vLLM replicas.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…
- `name` (string, required): Registered adapter name
- `replicas` (string, required): Comma-separated host:port list

### `ml_vllm_lora_unload` (~121 tokens)

Hot-unload a LoRA adapter from a running vLLM server.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…
- `endpoint` (string, required): vLLM endpoint
- `name` (string, required): Adapter name to unload

### `database_weaviate_create_collection` (~98 tokens)

Create a Weaviate collection.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…
- `name` (string, required): Collection name

### `database_weaviate_delete_collection` (~98 tokens)

Delete a Weaviate collection.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…
- `name` (string, required): Collection name

### `database_weaviate_hybrid_search` (~113 tokens)

Hybrid vector + BM25 search in a Weaviate collection.

Input parameters:

- `collection` (string, required): collection
- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…
- `query` (string, required): Text query

### `database_weaviate_insert` (~118 tokens)

Insert objects into a Weaviate collection.

Input parameters:

- `collection` (string, required): collection
- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…
- `objects` (string, required): JSON list of objects [{properties: {...}, vector: [...]}]

### `database_weaviate_list_collections` (~89 tokens)

List Weaviate collections.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…

### `database_weaviate_query` (~110 tokens)

Vector similarity search in a Weaviate collection.

Input parameters:

- `collection` (string, required): collection
- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…
- `vector` (string, required): JSON array of floats

### `database_weaviate_up` (~88 tokens)

Initialize a Weaviate connection.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…

### `train_unsloth_run` (~116 tokens)

Run an Unsloth local model server.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…
- `model` (string, required): Model to serve (e.g. unsloth/Llama-3.1-8B)

### `train_unsloth_start` (~103 tokens)

Start a coding agent backed by Unsloth's local model server.

Input parameters:

- `agent` (string, required): agent
- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…

### `train_unsloth_stop` (~96 tokens)

Stop a running Unsloth server started with `unsloth run`.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…

### `create_sqlite_connection` (~98 tokens)

Create a SQLite database connection with auto-table creation. Returns a connection ID for subsequent operations.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…

### `create_postgresql_connection` (~137 tokens)

Create a PostgreSQL database connection with auto-table creation. Returns a connection ID for subsequent operations.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…
- `database` (string, required): Database name
- `host` (string, required): PostgreSQL host address
- `password` (string, required): Database password
- `user` (string, required): Database user

### `query_database` (~119 tokens)

Execute a SELECT query on a database connection. Returns query results as a list of dictionaries.

Input parameters:

- `connection_id` (string, required): Database connection ID
- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…
- `sql` (string, required): SQL SELECT query to execute

### `upsert_database` (~137 tokens)

Insert or update data in a database table. Performs upsert operation (insert or update on conflict).

Input parameters:

- `connection_id` (string, required): Database connection ID
- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…
- `data` (object, required): Data to insert/update as key-value pairs
- `table` (string, required): Table name (without prefix)

### `get_database_connection` (~104 tokens)

Get information about a database connection including type, status, and configuration.

Input parameters:

- `connection_id` (string, required): Database connection ID
- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…

### `secrets_set` (~145 tokens)

Store a secret in the Terradev secrets store. Values are encrypted at rest.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…
- `key` (string, required): Credential key name (e.g. api_key, secret_key)
- `provider` (string, required): Cloud provider name (e.g. runpod, aws, vastai)
- `value` (string, required): Secret value to store

### `secrets_get` (~109 tokens)

Retrieve a stored secret. By default the value is masked.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…
- `key` (string, required): Credential key name
- `provider` (string, required): Provider name

### `secrets_list` (~90 tokens)

List stored provider and key names. Values are never shown.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…

### `secrets_remove` (~99 tokens)

Remove a provider or a single key from the secret store.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…
- `provider` (string, required): Provider name

### `secrets_sync` (~94 tokens)

Import TERRADEV_* environment variables into the secret store for supported cloud providers.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…

### `secrets_verify` (~90 tokens)

Check which providers are fully configured and which keys are missing.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…

### `secrets_env` (~102 tokens)

Print environment-style export lines for a provider. By default values are masked.

Input parameters:

- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…
- `provider` (string, required): Provider name

### `secrets_run` (~106 tokens)

Run a shell command with secrets injected into the environment.

Input parameters:

- `command` (string, required): Shell command to execute (will be split into arguments)
- `credentials` (object): BYOAPI — your own provider credentials as env vars for this call only (never stored). E.g. {"RUNPOD_API_KEY": "..."}, {"AWS_ACCESS_KEY_ID": "...", "AWS_SECRET_ACCESS_KEY": "..."}, or the TERRADEV_<PR…

## Diagnostics

Captured diagnostic sections: TLS, DNSSEC, Authorisation, Transports. The full working is on the page: https://verifymcp.io/servers/theoddden-terradev/terradev-mcp#diagnostics

## Score history

- 2026-09-21: 63
- 2026-09-20: 63
- 2026-09-19: 62
- 2026-09-18: 62
- 2026-09-17: 61
- 2026-09-16: 61
- 2026-09-15: 60
- 2026-09-14: 60
- 2026-09-13: 59
- 2026-09-12: 59
- 2026-09-11: 59
- 2026-09-10: 58
- 2026-09-09: 58
- 2026-09-08: 57

## Common questions

### What is the Terradev MCP server?

Terradev is an MCP server listed in the public MCP registry as io.github.theoddden/terradev. Cross-cloud GPU orchestration CLI. This page covers its hosted endpoint (https://terradev-mcp.terradev.cloud/sse).

### Is the Terradev MCP server safe to use?

Terradev scores 63 out of 100 on VerifyMCP. That is a record of what we were able to check automatically, not an endorsement. The category breakdown on this page shows every signal behind the number, including the ones we could not confirm.

### What tools does the Terradev MCP server expose?

Terradev exposes 245 tools: provision_gpu, preflight_report, preflight_gpu_check, preflight_network_check, local_scan, and 240 more. Their descriptions and schemas cost roughly 27,281 tokens of context every time the server is loaded.

### Does the Terradev MCP server require authentication?

No. We connected to Terradev without credentials and it answered, so anything it exposes is reachable by anyone who knows the address.

### Is the Terradev MCP server still maintained?

Terradev is still listed as active in the MCP registry. We last reached this channel on 21 September 2026. Those dates come from our own scans of the registry and the channel itself, not from anything the publisher announced.

## Links

- Remote endpoint: https://terradev-mcp.terradev.cloud/sse
- Changelog RSS feed: https://verifymcp.io/servers/theoddden-terradev/terradev-mcp.xml
- Changelog JSON feed: https://verifymcp.io/servers/theoddden-terradev/terradev-mcp.json
- HTML version of this page: https://verifymcp.io/servers/theoddden-terradev/terradev-mcp