# XFuel (npm · xfuel-mcp)

Submit AI inference to XFuel, pay per task (USDC via x402 or TFUEL), fetch/verify ZK proofs.

- Trust score: 67/100 (medium)
- Change this week: +24
- Registry status: active
- Liveness: live
- Owner verified: no
- Last scored: 2026-08-03

## Components

- npm · `xfuel-mcp`: 67/100 (this document), [markdown](https://verifymcp.io/servers/xfuel-lab-xfuel-mcp/xfuel-mcp.md), [page](https://verifymcp.io/servers/xfuel-lab-xfuel-mcp/xfuel-mcp)

## Channel facts

- Registry: `npm`
- Package: `xfuel-mcp`
- Version: `0.1.1`
- Transport: `stdio`

## Trust breakdown

How this component scores in each security and reliability category. Every signal is checked automatically from public evidence about the published package, including repeated runs of it in an isolated sandbox, 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 2026-08-03.

- **Supply Chain Security**: 88/100
  - No malware found by supply-chain analysis.
  - Only part of the dependency tree could be resolved (118 of 119), so this covers what we could see, not the whole tree.
  - No install/post-install scripts declared.
  - Only part of the dependency tree could be resolved (118 of 119), so this covers what we could see, not the whole tree.
- **Provenance & Transparency**: 45/100
  - Source repository is publicly reachable at the declared URL.
  - Provenance check failed: no build-provenance attestation is published.
  - Clear OSI-approved license (Apache-2.0).
  - Actively maintained (last published 24 days ago).
  - Disclosure check failed: no security disclosure policy was found in the source repository.
- **Schema Quality & AI Usability**: 61/100
  - AI-judged instruction clarity (good).
  - Context-footprint check failed: tool/resource definitions use about 1858 tokens (~232/item across 8 items; 8 tools + 0 resources), over budget; trim descriptions and params.
  - Usage-examples check failed: none of the tools include examples.
- **Stability & Change Management**: 27/100
  - Stability observed for 8 of 30 days with no destabilising changes; credit accrues until the full window elapses.
- **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.
- **Capabilities**: 100/100
  - Implements a supported MCP spec version (2025-11-25); the latest is 2026-07-28.

## Install

### Claude

```bash
claude mcp add xfuel-lab-xfuel-mcp -- npx -y xfuel-mcp
```

### Codex

```bash
codex mcp add xfuel-lab-xfuel-mcp -- npx -y xfuel-mcp
```

### opencode

```json
{
  "$schema": "https://opencode.ai/config.json",
  "mcp": {
    "xfuel-lab-xfuel-mcp": {
      "type": "local",
      "command": [
        "npx",
        "-y",
        "xfuel-mcp"
      ],
      "enabled": true
    }
  }
}
```

### OpenClaw

```bash
openclaw mcp add xfuel-lab-xfuel-mcp --command npx --arg -y --arg xfuel-mcp
```

### Hermes

```yaml
mcp_servers:
  xfuel-lab-xfuel-mcp:
    command: "npx"
    args: ["-y", "xfuel-mcp"]
```

### Other

```json
{
  "mcpServers": {
    "xfuel-lab-xfuel-mcp": {
      "command": "npx",
      "args": [
        "-y",
        "xfuel-mcp"
      ]
    }
  }
}
```

## 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.

### 2026-08-03 (score 67, +4)

- [functional improvement] Stability: unverified → 0.27

### 2026-08-02 (score 63, +58)

- [security regression] Provenance: unverified → fail
- [security improvement] Install scripts: unverified → pass
- [security improvement] Known CVEs: unverified → partial
- [security improvement] Malware scan: unverified → pass
- [security] Stability: Stability not yet verified: not enough scan history yet (needs a 30-day window).
- [functional regression] Security disclosure: fail → unverified
- [functional improvement] License: unverified → pass
- [functional improvement] Dependency health: unverified → partial
- [functional improvement] Maintenance: unverified → pass
- [functional improvement] MCP protocol: unverified → pass
- [functional improvement] Schema quality: unverified → good
- [functional improvement] Tool coverage: unverified → 100
- [functional] Licence: Apache-2.0

### 2026-08-01 (score 5, −13)

- [functional regression] Tool coverage: 100 → unverified
- [functional] First check of Schema quality: unverified

### 2026-07-31 (score 18, −7)

- [functional] We updated how we score, so this day's move reflects our rubric, not a change to the server

### 2026-07-30 (score 25, −18)

- [security regression] Malware scan: pass → unverified

### 2026-07-27 (score 43)

First indexed and scored.

## MCP tools (8)

### `submit_inference` (~455 tokens)

Submit AI inference task

Submit an AI inference task to the XFuel Protocol. XFuel routes it to a GPU provider (Theta EdgeCloud → DePIN fallbacks), and settles with a ZK proof.

Args:
  \- model (string): model id, e.g. "llama-3-70b"
  \- sender (string): the 0x address that owns/pays for the task
  \- amount (string): gross task value in the smallest unit (wei); minimum 10000
  \- chain_id ('theta'|'bittensor'|'akash'|'osmosis'|'persistence'): settlement network (default 'theta')
  \- input_hash (string, optional): keccak256 of your input (recommended for inference)
  \- memo (string, optional): free-form note echoed on the task
  \- max_gpu_hours (string, optional): compute budget hint
  \- subnet_id (number, optional): Bittensor subnet id when chain_id='bittensor'
  \- callback_url (string, optional): webhook that receives a signed TaskSettled event

Returns JSON: { task_id, status, payment_rail, fee_bps, gross_amount, fee_amount, net_amount, links }.
Poll progress with get_task_status(task_id); fetch settlement with get_proof(task_id).

Note: this submits with the server's default (unpaid/TFUEL) rail. For USDC/x402
settlement (which needs an agent-side signer) use the xfuel-sdk directly.

Input parameters:

- `amount` (string, required): Gross task value in smallest unit (wei); min 10000
- `callback_url` (string): Webhook for signed TaskSettled event
- `chain_id` (string): Settlement network (default theta)
- `input_hash` (string): keccak256 of your input
- `max_gpu_hours` (string): Compute budget hint
- `memo` (string): Free-form note
- `model` (string, required): Model id, e.g. "llama-3-70b"
- `sender` (string, required): 0x address that owns/pays for the task
- `subnet_id` (integer): Bittensor subnet id

### `pay_with_usdc` (~556 tokens)

Submit + pay for inference with USDC (x402)

Submit an AI inference task and settle it with USDC via x402 (the default XFuel rail).
The server signs an EIP-3009 USDC authorization on Base with its configured payer wallet, so
the agent doesn't manage the handshake. If the server has x402 disabled it transparently
falls back to the TFUEL rail (no payment is made).

Requires the server to be started with XFUEL_PAYER_PRIVATE_KEY. If it is not set, this tool
returns a clear "not configured" message — use submit_inference (TFUEL) instead, or the
xfuel-sdk with your own payer. The USDC network is chosen by the server's x402 challenge
(the payer signs for whatever network the challenge specifies, e.g. Base or Base Sepolia).

Args:
  \- model (string): model id, e.g. "llama-3-70b"
  \- amount (string): gross task value in the smallest unit (wei); minimum 10000
  \- sender (string, optional): the 0x address that owns the task (default: the payer wallet address)
  \- chain_id ('theta'|'bittensor'|'akash'|'osmosis'|'persistence'): settlement network (default 'theta')
  \- input_hash (string, optional): keccak256 of your input (recommended for inference)
  \- memo (string, optional): free-form note echoed on the task
  \- max_gpu_hours (string, optional): compute budget hint
  \- subnet_id (number, optional): Bittensor subnet id when chain_id='bittensor'
  \- callback_url (string, optional): webhook that receives a signed TaskSettled event

Returns JSON: { task_id, status, payment_rail, fee_bps, gross_amount, fee_amount, net_amount, links }.
'payment_rail' is 'usdc' when the x402 handshake ran, or 'tfuel' if the server fell back.

Input parameters:

- `amount` (string, required): Gross task value in smallest unit (wei); min 10000
- `callback_url` (string): Webhook for signed TaskSettled event
- `chain_id` (string): Settlement network (default theta)
- `input_hash` (string): keccak256 of your input
- `max_gpu_hours` (string): Compute budget hint
- `memo` (string): Free-form note
- `model` (string, required): Model id, e.g. "llama-3-70b"
- `sender` (string): 0x address that owns the task (default: payer wallet)
- `subnet_id` (integer): Bittensor subnet id

### `get_task_status` (~155 tokens)

Get task status

Get the current status of an XFuel task, including proof outcome and fee breakdown.

Args:
  \- task_id (string): the id returned by submit_inference

Returns JSON: { task_id, status, proof_outcome, message_type, chain_id, gross_amount,
fee_amount, net_amount, fee_bps, payment_rail, payment_ref, result, sp1_proof, created_at, updated_at }.
'status' reaches a terminal value ('completed' | 'fee_collected' | 'failed'); 'proof_outcome'
is one of 'pending' | 'valid' | 'regenerable' | 'invalid'.

Input parameters:

- `task_id` (string, required): Task id from submit_inference

### `get_proof` (~145 tokens)

Get ZK settlement proof

Fetch the SP1 ZK settlement proof for a settled task.

Args:
  \- task_id (string): the id returned by submit_inference

Returns JSON: { task_id, status, proof_outcome, payment_binding, sp1_proof: { proof,
publicInputs, nullifier, provingTimeMs }, fee }.

The proof attests settlement metadata + a commitment to the output hash (NOT inference
correctness). To validate it, use verify_proof(task_id). Fails if the task has not
settled yet — poll get_task_status until proof_outcome is 'valid'.

Input parameters:

- `task_id` (string, required): Task id from submit_inference

### `verify_proof` (~242 tokens)

Verify a task proof

Independently verify a settled task's proof client-side (the "prove it" flow in one call):
  1\. a proof is present and proof_outcome === 'valid';
  2\. the x402 payment binding (if present) re-derives to the committed value;
  3\. (optional) the on-chain nullifier's spent state, when check_nullifier=true AND the
     server was started with XFUEL_RPC_URL + ZK_VERIFIER_ADDRESS.

Args:
  \- task_id (string): the id returned by submit_inference
  \- check_nullifier (boolean, optional): also read the on-chain nullifier state (default false)

Returns JSON: { ok, checks: { hasProof, proofOutcomeValid, paymentBinding, nullifier }, reasons }.
'ok' is true when the proof is present + valid and any present payment binding is consistent.
The nullifier read is informational and does not gate 'ok'.

Input parameters:

- `check_nullifier` (boolean): Also read the on-chain nullifier state (needs RPC + zkVerifier configured)
- `task_id` (string, required): Task id from submit_inference

### `quote_task` (~150 tokens)

Quote / price a task

Preview per-rail pricing for a task WITHOUT creating it (no side effects). Use this
before submit_inference to show the user cost across payment rails.

Args:
  \- model_id (string, optional): model id to price (some models have overrides)
  \- amount (string, optional): TFUEL task value in wei (echoed back in the tfuel rail)

Returns JSON: { recommended, default_rail, rails: { usdc: { enabled, asset, network,
decimals, amount, pay_to }, tfuel: { amount } } }.

Input parameters:

- `amount` (string): TFUEL task value in wei
- `model_id` (string): Model id to price

### `get_health` (~67 tokens)

Get XFuel API health

Check the connected XFuel API's health and configuration (fee config, supported
chains, message types, demo limits). Useful for discovery/diagnostics.

Args: none.

Returns JSON: the /health payload (status, server, version, fee_config, chains, message_types, …).

### `list_models` (~88 tokens)

List routable models

List the models XFuel can route inference to (OpenAI-compatible GET /v1/models).
Call this first to discover valid model ids, then pass one as 'model' to submit_inference
or pay_with_usdc. No side effects.

Args: none.

Returns JSON: { object: 'list', data: [{ id, object: 'model', created, owned_by }] }.

## Diagnostics

Captured diagnostic sections: Provenance, Dependencies. The full working is on the page: https://verifymcp.io/servers/xfuel-lab-xfuel-mcp/xfuel-mcp#diagnostics

## Score history

- 2026-08-03: 67
- 2026-08-02: 63
- 2026-08-01: 5
- 2026-07-31: 18
- 2026-07-30: 25
- 2026-07-28: 43
- 2026-07-27: 43

## Links

- npm package: https://www.npmjs.com/package/xfuel-mcp
- Socket report: https://socket.dev/npm/package/xfuel-mcp
- Repository: https://github.com/XFuel-Lab/xfuel-protocol
- Website: https://github.com/XFuel-Lab/xfuel-protocol/tree/main/xfuel-mcp#readme
- Changelog RSS feed: https://verifymcp.io/servers/xfuel-lab-xfuel-mcp/xfuel-mcp/changelog.xml
- Changelog JSON feed: https://verifymcp.io/servers/xfuel-lab-xfuel-mcp/xfuel-mcp/changelog.json
- HTML version of this page: https://verifymcp.io/servers/xfuel-lab-xfuel-mcp/xfuel-mcp