# LLM CLI Gateway (npm · llm-cli-gateway)

One MCP endpoint for Claude Code, Codex, Gemini, Grok and Mistral CLIs, with durable async jobs.

- Trust score: 71/100 (medium)
- Change this week: +30
- Registry status: active
- Liveness: live
- Owner verified: no
- Last scored: 2026-08-03

## Components

- npm · `llm-cli-gateway`: 71/100 (this document), [markdown](https://verifymcp.io/servers/verivus-oss-llm-cli-gateway/llm-cli-gateway.md), [page](https://verifymcp.io/servers/verivus-oss-llm-cli-gateway/llm-cli-gateway)

## Channel facts

- Registry: `npm`
- Package: `llm-cli-gateway`
- Version: `1.17.1`
- 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**: 87/100
  - No malware found by supply-chain analysis.
  - Only part of the dependency tree could be resolved (132 of 136), 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 (132 of 136), so this covers what we could see, not the whole tree.
- **Provenance & Transparency**: 97/100
  - Source repository is publicly reachable at the declared URL.
  - Cryptographically verified build provenance (signed, bound to verivus-oss/llm-cli-gateway).
  - Clear OSI-approved license (MIT).
  - Actively maintained (last published 6 days ago).
  - Disclosure check failed: no security disclosure policy was found in the source repository.
- **Schema Quality & AI Usability**: 70/100
  - 100% of prompts and resources have a non-trivial description (not blank, and not just the item's name).
  - AI-judged instruction clarity (fair).
  - Context-footprint check failed: tool/resource definitions use about 6187 tokens (~147/item across 42 items; 28 tools + 14 resources), over budget; trim descriptions and params.
  - Usage-examples check failed: none of the tools include examples.
- **Stability & Change Management**: 23/100
  - Stability observed for 7 of 30 days with no destabilising changes; credit accrues until the full window elapses.
- **Tool Coverage**: 33/100
  - 0% 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 verivus-oss-llm-cli-gateway -- npx -y llm-cli-gateway
```

### Codex

```bash
codex mcp add verivus-oss-llm-cli-gateway -- npx -y llm-cli-gateway
```

### opencode

```json
{
  "$schema": "https://opencode.ai/config.json",
  "mcp": {
    "verivus-oss-llm-cli-gateway": {
      "type": "local",
      "command": [
        "npx",
        "-y",
        "llm-cli-gateway"
      ],
      "enabled": true
    }
  }
}
```

### OpenClaw

```bash
openclaw mcp add verivus-oss-llm-cli-gateway --command npx --arg -y --arg llm-cli-gateway
```

### Hermes

```yaml
mcp_servers:
  verivus-oss-llm-cli-gateway:
    command: "npx"
    args: ["-y", "llm-cli-gateway"]
```

### Other

```json
{
  "mcpServers": {
    "verivus-oss-llm-cli-gateway": {
      "command": "npx",
      "args": [
        "-y",
        "llm-cli-gateway"
      ]
    }
  }
}
```

## 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 71, +3)

- [functional improvement] Stability: unverified → 0.23

### 2026-08-02 (score 68, +50)

- [security improvement] Install scripts: unverified → pass
- [security improvement] Provenance: 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).
- [security] The attested source repository moved: verivus-oss/llm-cli-gateway
- [functional regression] Security disclosure: fail → unverified
- [functional improvement] License: unverified → pass
- [functional improvement] Schema quality: unverified → fair
- [functional improvement] Dependency health: unverified → partial
- [functional improvement] Maintenance: unverified → pass
- [functional improvement] MCP protocol: unverified → pass
- [functional] Licence: MIT

### 2026-07-31 (score 18, −5)

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

### 2026-07-30 (score 23, −18)

- [security regression] Malware scan: pass → unverified

### 2026-07-27 (score 41)

First indexed and scored.

## MCP tools (28)

### `validate_with_models` (~63 tokens)

Input parameters:

- `focus` (string): What reviewers should pay attention to.
- `judgeModel`: Optional provider to run an explicit judge synthesis job.
- `models` (array): Providers to ask. Defaults to Claude and Codex.
- `question` (string, required): Question or content to validate.

### `second_opinion` (~42 tokens)

Input parameters:

- `answer` (string, required): Answer to review.
- `model` (string): Provider to ask for the second opinion.
- `question` (string): Original question, if available.

### `compare_answers` (~30 tokens)

Input parameters:

- `answers` (array, required): Two or more answers to compare.
- `question` (string, required): Question the answers respond to.

### `red_team_review` (~47 tokens)

Input parameters:

- `content` (string, required): Plan, answer, or document to challenge.
- `models` (array): Providers to ask for adversarial review.
- `riskLevel` (string): How aggressively to review.

### `consensus_check` (~31 tokens)

Input parameters:

- `claim` (string, required): Claim to check across providers.
- `models` (array): Providers to ask for agreement or disagreement.

### `ask_model` (~26 tokens)

Input parameters:

- `model` (string): Provider to ask.
- `question` (string, required): Question for one provider.

### `synthesize_validation` (~45 tokens)

Input parameters:

- `judgeModel`: Provider to run the judge synthesis.
- `providerResults` (array, required): Terminal normalized provider results from job_result.
- `question` (string, required): Original request that was validated.

### `list_available_models` (~6 tokens)

### `job_status` (~16 tokens)

Input parameters:

- `jobId` (string, required): Validation job ID.

### `job_result` (~45 tokens)

Input parameters:

- `jobId` (string, required): Validation job ID.
- `maxChars` (integer): Maximum result size.
- `provider` (string): Provider that produced the job, used for normalized validation output.

### `claude_request` (~1118 tokens)

Input parameters:

- `addDir` (array): Claude --add-dir: additional directories the CLI is allowed to read/write beyond the process cwd. Each entry is emitted as its own --add-dir instance.
- `agent` (string): Claude --agent: dispatch to a named single sub-agent.
- `agents` (object): Claude --agents: inline JSON map of agent name → { description, prompt, tools?, model? }.
- `allowedTools` (array): Allowed tools (['Bash(git:*)','Edit','Write'])
- `appendSystemPrompt` (string): Claude --append-system-prompt: append to the existing system prompt.
- `approvalPolicy` (string): Approval policy override
- `approvalStrategy` (string): Approval strategy
- `continueSession` (boolean): Continue active session
- `correlationId` (string): Request trace ID (auto if omitted)
- `createNewSession` (boolean): Force new session
- `dangerouslySkipPermissions` (boolean): DEPRECATED: prefer `permissionMode: "bypassPermissions"`. Maps to it when `permissionMode` is unset.
- `disallowedTools` (array): Disallowed tools
- `effort` (string): Claude --effort: low|medium|high|xhigh|max.
- `excludeDynamicSystemPromptSections` (boolean): Claude --exclude-dynamic-system-prompt-sections: trim dynamic context blocks from the system prompt.
- `fallbackModel` (string): Claude --fallback-model: model name to auto-fallback to when the default model is overloaded (effective only with --print, which the gateway always uses).
- `forceRefresh` (boolean): Bypass dedup and force a fresh CLI run even if a recent identical request exists
- `forkSession` (boolean): Claude --fork-session: branch from an existing session into a fresh fork.
- `idleTimeoutMs` (integer): Idle timeout in ms (min 30s, max 1h, omit=CLI default)
- `jsonSchema`: Claude --json-schema: JSON Schema literal (NOT a path) constraining structured output. Object values are JSON.stringify-d; string values are passed verbatim. Use with outputFormat='json'.
- `maxBudgetUsd` (number): Claude --max-budget-usd: spend cap for this request in USD.
- `maxTurns` (integer): Claude --max-turns: cap on agent loop iterations.
- `mcpServers` (array): MCP servers exposed to Claude
- `model` (string): Model name or alias (e.g. sonnet, claude-sonnet-4-5-20250929, latest)
- `optimizePrompt` (boolean): Optimize prompt before execution
- `optimizeResponse` (boolean): Optimize response output
- `outputFormat` (string): Output format (text|json|stream-json). DEFAULT: stream-json — the gateway parses NDJSON usage events to extract input/output/cache_read/cache_creation tokens + cost + model, persists them to the flig…
- `permissionMode` (string): Claude --permission-mode: default|acceptEdits|plan|auto|dontAsk|bypassPermissions. `default` is a no-op (no flag emitted).
- `prompt` (string): Prompt text for Claude (mutually exclusive with promptParts)
- `promptParts` (object): Cache-aware structured prompt: { system?, tools?, context?, task, cacheControl? }. Use for repeated calls that share a stable prefix — `system`/`tools`/`context` are the stable head; `task` is the vo…
- `sessionId` (string): Session ID (uses active if omitted)
- `strictMcpConfig` (boolean): Restrict Claude to provided MCP config only
- `systemPrompt` (string): Claude --system-prompt: replace the system prompt entirely.
- `worktree`: Slice λ: run this request inside a dedicated git worktree owned by the gateway. `true` creates a fresh worktree at `<repoRoot>/.worktrees/<uuid>` branched from HEAD. `{ name?, ref? }` lets the caller…

### `codex_request` (~1014 tokens)

Input parameters:

- `addDir` (array): Codex --add-dir <DIR>: additional writable workspace directories. Emitted once per entry on new sessions only; resume inherits the original session's writable-dir policy.
- `approvalPolicy` (string): Approval policy override
- `approvalStrategy` (string): Approval strategy
- `askForApproval` (string): Codex --ask-for-approval: untrusted|on-request|never.
- `configOverrides` (object): Codex -c key=value overrides. Keys: /^[a-zA-Z0-9._]+$/. Values: no CR/LF.
- `correlationId` (string): Request trace ID (auto if omitted)
- `createNewSession` (boolean): Force a fresh session (no resume)
- `dangerouslyBypassApprovalsAndSandbox` (boolean): Run Codex without approvals/sandbox
- `ephemeral` (boolean): Codex --ephemeral: do not persist the session to disk.
- `forceRefresh` (boolean): Bypass dedup and force a fresh CLI run even if a recent identical request exists
- `fullAuto` (boolean): DEPRECATED: prefer `sandboxMode` + `askForApproval`. Expands to `--sandbox workspace-write --ask-for-approval never`.
- `idleTimeoutMs` (integer): Idle timeout in ms (min 30s, max 1h, omit=CLI default)
- `ignoreRules` (boolean): Codex --ignore-rules: skip project rule files for this run.
- `ignoreUserConfig` (boolean): Codex --ignore-user-config: ignore ~/.codex/config.toml for this run.
- `images` (array): Codex -i <path>: image attachments. Each path must exist; missing paths fail fast.
- `mcpServers` (array): MCP server names for approval tracking (Codex manages its own MCP config)
- `model` (string): Model name or alias (e.g. gpt-5.4, latest)
- `optimizePrompt` (boolean): Optimize prompt before execution
- `optimizeResponse` (boolean): Optimize response output
- `outputFormat` (string): Codex output format. `json` emits --json (JSONL events) so token usage and cost are parsed and reported in the flight recorder. `text` is the default.
- `outputSchema`: Codex --output-schema. Pass a path (string) or an inline JSON Schema object; object is materialised to a 0o600 temp file under os.tmpdir() and deleted after the run.
- `profile` (string): Codex --profile <name>: select a profile from ~/.codex/config.toml.
- `prompt` (string): Prompt text for Codex (mutually exclusive with promptParts)
- `promptParts` (object): Cache-aware structured prompt: { system?, tools?, context?, task }. Mutually exclusive with prompt. Stable parts hash into cache_state for prefix-discipline tracking.
- `resumeLatest` (boolean): Resume the most recent Codex session in the current cwd via `codex exec resume --last`. Ignored if sessionId is set.
- `sandboxMode` (string): Codex --sandbox: read-only|workspace-write|danger-full-access.
- `search` (boolean): Emit Codex --search to enable web search.
- `sessionId` (string): Codex session UUID to resume via `codex exec resume <ID>`. Must be a real Codex session ID (from `~/.codex/sessions/` or the `codex resume` picker). Gateway-generated `gw-*` IDs are rejected.
- `useLegacyFullAutoFlag` (boolean): Escape hatch: emit `--full-auto` directly instead of expanding (deprecated).
- `workingDir` (string): Codex -C/--cd <DIR>: working root for this session. Emitted on new sessions only; resume inherits the original session's cwd via CODEX_RESUME_FILTERED_FLAGS.
- `worktree`: Slice λ: run this request inside a dedicated git worktree owned by the gateway. `true` creates a fresh worktree at `<repoRoot>/.worktrees/<uuid>` branched from HEAD. `{ name?, ref? }` lets the caller…

### `codex_fork_session` (~183 tokens)

Input parameters:

- `askForApproval` (string): Codex --ask-for-approval: untrusted|on-request|never.
- `correlationId` (string): Request trace ID (auto if omitted)
- `forkLast` (boolean): Fork from the most recent Codex session. Mutually exclusive with `sessionId`.
- `idleTimeoutMs` (integer): Idle timeout in ms (min 30s, max 1h, omit=CLI default)
- `model` (string): Model name or alias (e.g. gpt-5.5, latest)
- `prompt` (string, required): Prompt text for the forked Codex session
- `sandboxMode` (string): Codex --sandbox: read-only|workspace-write|danger-full-access.
- `sessionId` (string): Codex session UUID to fork from. Mutually exclusive with `forkLast`.

### `gemini_request` (~728 tokens)

Input parameters:

- `adminPolicyFiles` (array): Admin policy file paths (--admin-policy <path>, one per file). Paths must exist.
- `allowedTools` (array): Allowed tools (['Write','Edit','Bash'])
- `approvalMode` (string): Approval: default|auto_edit|yolo|plan
- `approvalPolicy` (string): Approval policy override
- `approvalStrategy` (string): Approval strategy
- `attachments` (array): Absolute file paths prepended as @<path> tokens to the prompt
- `correlationId` (string): Request trace ID (auto if omitted)
- `createNewSession` (boolean): Force new session
- `forceRefresh` (boolean): Bypass dedup and force a fresh CLI run even if a recent identical request exists
- `idleTimeoutMs` (integer): Idle timeout in ms (min 30s, max 1h, omit=CLI default)
- `includeDirs` (array): Additional workspace directories
- `mcpServers` (array): MCP server names passed to Gemini as --allowed-mcp-server-names
- `model` (string): Model name or alias (e.g. gemini-3-pro-preview, gemini-2.5-flash, pro, flash, latest)
- `optimizePrompt` (boolean): Optimize prompt before execution
- `optimizeResponse` (boolean): Optimize response output
- `outputFormat` (string): Gemini output format. `json` emits `-o json` (single JSON with usageMetadata). `stream-json` emits `-o stream-json` (NDJSON event stream — `init`/`message`/`result` lines, usage extracted from the te…
- `policyFiles` (array): Policy file paths (--policy <path>, one per file). Paths must exist.
- `prompt` (string): Prompt text for Gemini (mutually exclusive with promptParts)
- `promptParts` (object): Cache-aware structured prompt: { system?, tools?, context?, task }. Mutually exclusive with prompt. Stable parts hash into cache_state for prefix-discipline tracking.
- `resumeLatest` (boolean): Resume latest session
- `sandbox` (boolean): Run Gemini in sandbox mode (-s)
- `sessionId` (string): Session ID or 'latest'
- `skipTrust` (boolean): Emit `--skip-trust` so Gemini trusts the workspace for this session and skips the interactive trust prompt (Phase 4 slice γ). Required for headless runs in fresh workspaces.
- `worktree`: Slice λ: run this request inside a dedicated git worktree owned by the gateway. `true` creates a fresh worktree at `<repoRoot>/.worktrees/<uuid>` branched from HEAD. `{ name?, ref? }` lets the caller…

### `grok_request` (~943 tokens)

Input parameters:

- `allow` (array): Grok --allow <RULE>: permission allow rules. Each entry is emitted as its own --allow instance (per `grok --help`: "Repeat to add multiple rules").
- `allowedTools` (array): Allowed built-in tools (passed as --tools comma list)
- `alwaysApprove` (boolean): Auto-approve all tool executions (--always-approve)
- `approvalPolicy` (string): Approval policy override
- `approvalStrategy` (string): Approval strategy
- `correlationId` (string): Request trace ID (auto if omitted)
- `createNewSession` (boolean): Force new session
- `deny` (array): Grok --deny <RULE>: permission deny rules. Each entry is emitted as its own --deny instance (per `grok --help`: "Repeat to add multiple rules").
- `disallowedTools` (array): Disallowed built-in tools (passed as --disallowed-tools comma list)
- `effort` (string): Grok effort level
- `forceRefresh` (boolean): Bypass dedup and force a fresh CLI run even if a recent identical request exists
- `idleTimeoutMs` (integer): Idle timeout in ms (min 30s, max 1h, omit=CLI default)
- `maxTurns` (integer): Grok `--max-turns N`: cap on agent-loop iterations for cost / latency control (Phase 4 slice δ). Bounded to safe integers ≤ 10000.
- `mcpServers` (array): MCP server names for approval tracking (Grok manages its own MCP config via `grok mcp`)
- `model` (string): Model name or alias (e.g. grok-build, latest)
- `optimizePrompt` (boolean): Optimize prompt before execution
- `optimizeResponse` (boolean): Optimize response output
- `outputFormat` (string): Output format (plain|json|streaming-json). Grok default is plain.
- `permissionMode` (string): Grok permission mode
- `prompt` (string): Prompt text for Grok (mutually exclusive with promptParts)
- `promptParts` (object): Cache-aware structured prompt: { system?, tools?, context?, task }. Mutually exclusive with prompt. Stable parts hash into cache_state for prefix-discipline tracking.
- `reasoningEffort` (string): Reasoning effort for reasoning models
- `resumeLatest` (boolean): Resume most recent Grok session in cwd (--continue)
- `rules` (string): Grok --rules <RULES>: extra rules to append to the system prompt. Supports `@file` prefix per `grok --help` to load from a file; gateway passes the value verbatim and lets Grok parse the prefix.
- `sandbox` (string): Grok --sandbox <PROFILE>: sandbox profile for filesystem and network access. Freeform per `grok --help` (no enum constraint on Grok 0.1.210); also settable via GROK_SANDBOX env var. Caller responsibi…
- `sessionId` (string): Session ID (user-provided CLI handle for --resume)
- `systemPromptOverride` (string): Grok --system-prompt-override <PROMPT>: replace the agent's system prompt entirely. Distinct from Claude's --system-prompt / --append-system-prompt (Grok has only one override flag, not a pair).
- `workingDir` (string): Grok --cwd <DIR>: working directory for this invocation. Lets headless callers run Grok against a directory other than the gateway process's cwd.
- `worktree`: Slice λ: run this request inside a dedicated git worktree owned by the gateway. `true` creates a fresh worktree at `<repoRoot>/.worktrees/<uuid>` branched from HEAD. `{ name?, ref? }` lets the caller…

### `mistral_request` (~943 tokens)

Input parameters:

- `addDir` (array): Vibe --add-dir <DIR>: additional writable workspace directories. Each entry is emitted as its own --add-dir instance (Vibe states this flag may be specified multiple times).
- `allowedTools` (array): Allowlist of built-in tools — each emitted as a separate --enabled-tools <tool> flag
- `approvalPolicy` (string): Approval policy override
- `approvalStrategy` (string): Approval strategy
- `correlationId` (string): Request trace ID (auto if omitted)
- `createNewSession` (boolean): Force new session
- `disallowedTools` (array): Accepted for caller parity; Vibe has no deny-list flag, so values are ignored (a warning is logged).
- `effort` (string): Vibe effort level
- `forceRefresh` (boolean): Bypass dedup and force a fresh CLI run even if a recent identical request exists
- `idleTimeoutMs` (integer): Idle timeout in ms (min 30s, max 1h, omit=CLI default)
- `maxPrice` (number): Vibe `--max-price DOLLARS`: interrupt the session when cumulative cost crosses this cap (programmatic mode only, Phase 4 slice δ). Bounded to finite values ≤ 10000 USD.
- `maxTokens` (integer): Vibe `--max-tokens N`: cap cumulative prompt + completion tokens for the session (programmatic mode only). Bounded to safe integers ≤ 100000000.
- `maxTurns` (integer): Vibe `--max-turns N`: cap the agent-loop iteration count (programmatic mode only, Phase 4 slice δ). Bounded to safe integers ≤ 10000.
- `mcpServers` (array): MCP server names for approval tracking (Vibe manages its own MCP config via `vibe mcp`)
- `model` (string): Model alias (e.g. mistral-medium-3.5, latest). Resolved alias is injected via VIBE_ACTIVE_MODEL env var; Vibe has no --model flag.
- `optimizePrompt` (boolean): Optimize prompt before execution
- `optimizeResponse` (boolean): Optimize response output
- `outputFormat` (string): Output format for Vibe 2.x (text|json|streaming). Legacy aliases plain→text and stream-json→streaming are accepted.
- `permissionMode` (string): Vibe agent mode (default|plan|accept-edits|auto-approve|chat|explore|lean). Defaults to auto-approve for programmatic use.
- `prompt` (string): Prompt text for Mistral Vibe (mutually exclusive with promptParts)
- `promptParts` (object): Cache-aware structured prompt: { system?, tools?, context?, task }. Mutually exclusive with prompt. Stable parts hash into cache_state for prefix-discipline tracking.
- `reasoningEffort` (string): Reasoning effort for reasoning models
- `resumeLatest` (boolean): Resume most recent Vibe session in cwd (--continue)
- `sessionId` (string): Session ID (user-provided CLI handle for --resume). Current Vibe defaults session logging on; doctor flags explicit [session_logging] enabled = false.
- `trust` (boolean): Emit `--trust` so Vibe trusts the cwd for this invocation only (not persisted to trusted_folders.toml) and skips the interactive trust prompt (Phase 4 slice γ).
- `workingDir` (string): Vibe --workdir <DIR>: change to this directory before running. Single value (Vibe accepts one --workdir per invocation).
- `worktree`: Slice λ: run this request inside a dedicated git worktree owned by the gateway. `true` creates a fresh worktree at `<repoRoot>/.worktrees/<uuid>` branched from HEAD. `{ name?, ref? }` lets the caller…

### `llm_process_health` (~7 tokens)

### `approval_list` (~27 tokens)

Input parameters:

- `cli` (string): Optional CLI filter
- `limit` (integer): Max number of approval records

### `list_models` (~30 tokens)

Input parameters:

- `cli` (string): CLI filter (claude|codex|gemini|grok|mistral)

### `cli_versions` (~30 tokens)

Input parameters:

- `cli` (string): CLI filter (claude|codex|gemini|grok|mistral)

### `upstream_contracts` (~51 tokens)

Input parameters:

- `cli` (string): CLI filter (claude|codex|gemini|grok|mistral)
- `probeInstalled` (boolean): When true, run local --help probes and compare advertised flags

### `cli_upgrade` (~67 tokens)

Input parameters:

- `cli` (string, required): CLI to upgrade
- `dryRun` (boolean): When true, return the upgrade plan without running it
- `target` (string): Package tag/version/target to install (default: latest)
- `timeoutMs` (integer): Upgrade timeout in ms when dryRun=false

### `session_create` (~52 tokens)

Input parameters:

- `cli` (string, required): CLI type (claude|codex|gemini|grok|mistral)
- `description` (string): Session description
- `setAsActive` (boolean): Set as active session

### `session_list` (~30 tokens)

Input parameters:

- `cli` (string): CLI filter (claude|codex|gemini|grok|mistral)

### `session_set_active` (~46 tokens)

Input parameters:

- `cli` (string, required): CLI type (claude|codex|gemini|grok|mistral)
- `sessionId` (string|null, required): Session ID (null to clear)

### `session_delete` (~15 tokens)

Input parameters:

- `sessionId` (string, required): Session ID

### `session_get` (~15 tokens)

Input parameters:

- `sessionId` (string, required): Session ID

### `session_clear_all` (~31 tokens)

Input parameters:

- `cli` (string): CLI filter (claude|codex|gemini|grok|mistral)

## Diagnostics

Captured diagnostic sections: Provenance, Dependencies. The full working is on the page: https://verifymcp.io/servers/verivus-oss-llm-cli-gateway/llm-cli-gateway#diagnostics

## Score history

- 2026-08-03: 71
- 2026-08-02: 68
- 2026-08-01: 18
- 2026-07-31: 18
- 2026-07-30: 23
- 2026-07-28: 41
- 2026-07-27: 41

## Links

- npm package: https://www.npmjs.com/package/llm-cli-gateway
- Socket report: https://socket.dev/npm/package/llm-cli-gateway
- Repository: https://github.com/verivus-oss/llm-cli-gateway
- Changelog RSS feed: https://verifymcp.io/servers/verivus-oss-llm-cli-gateway/llm-cli-gateway/changelog.xml
- Changelog JSON feed: https://verifymcp.io/servers/verivus-oss-llm-cli-gateway/llm-cli-gateway/changelog.json
- HTML version of this page: https://verifymcp.io/servers/verivus-oss-llm-cli-gateway/llm-cli-gateway