# ContextLLM API Documentation > ContextLLM — deep research & planning for coding. Your agent can ask for the latest technology over our ContextLLM MCP for deep research and planning. --- ## MCP (JSON-RPC 2.0) **Endpoint:** `POST https://trucopilot.com/api/contextllm/mcp` **Auth:** Bearer token in the `Authorization` header. Tokens are prefixed `cllm_*` and are minted from the Account → ContextLLM page. **Transport:** MCP Streamable HTTP (JSON-RPC 2.0 over HTTP POST). Compatible with Claude Code, Cursor, and any MCP client. ### Claude Code config ```json { "mcpServers": { "trucopilot-contextllm": { "url": "https://trucopilot.com/api/contextllm/mcp", "headers": { "Authorization": "Bearer cllm_YOUR_TOKEN" } } } } ``` --- ## MCP Tools ### resolve-library Find a repository by library or project name. Returns canonical owner/repo, stars, language, and description. **Input** | Field | Type | Required | Description | |-------|--------|----------|--------------------------------------------------| | name | string | yes | Library or project name (e.g. `"react"`, `"tokio"`) | **Request** ```json { "jsonrpc": "2.0", "id": 1, "method": "tools/call", "params": { "name": "resolve-library", "arguments": { "name": "tokio" } } } ``` **Response** ```json { "jsonrpc": "2.0", "id": 1, "result": { "content": [{ "type": "text", "text": "{\"full_name\":\"tokio-rs/tokio\",\"stars\":27000,\"language\":\"Rust\",\"description\":\"A runtime for writing reliable async Rust applications.\"}" }] } } ``` --- ### get-docs Retrieve a repo's summary, README excerpt, and top content chunks for a given `owner/repo`. **Input** | Field | Type | Required | Description | |-----------|--------|----------|-----------------------------------------------------| | full_name | string | yes | Repository in `owner/repo` form (e.g. `"tokio-rs/tokio"`) | **Request** ```json { "jsonrpc": "2.0", "id": 2, "method": "tools/call", "params": { "name": "get-docs", "arguments": { "full_name": "tokio-rs/tokio" } } } ``` **Response** ```json { "jsonrpc": "2.0", "id": 2, "result": { "content": [{ "type": "text", "text": "{\"summary_md\":\"Tokio is an async runtime for Rust ...\",\"readme_url\":\"https://github.com/tokio-rs/tokio/blob/master/README.md\",\"chunks\":[\"Tokio uses an epoll-based I/O driver ...\",\"The scheduler is work-stealing ...\"]}" }] } } ``` --- ### ask Deep research question answered over the full indexed corpus. Uses RAG: retrieve → rerank → synthesize. Returns a cited Markdown answer. **Input** | Field | Type | Required | Description | |---------|--------|----------|------------------------------------| | message | string | yes | Research question in natural language | **Request** ```json { "jsonrpc": "2.0", "id": 3, "method": "tools/call", "params": { "name": "ask", "arguments": { "message": "How does Tokio schedule async tasks?" } } } ``` **Response** ```json { "jsonrpc": "2.0", "id": 3, "result": { "content": [{ "type": "text", "text": "{\"answer_md\":\"Tokio uses a work-stealing scheduler ...\",\"citations\":[{\"full_name\":\"tokio-rs/tokio\",\"html_url\":\"https://github.com/tokio-rs/tokio\"}]}" }] } } ``` --- ### plan Generate a coding plan and recommended repo picks for a build goal. Returns structured Markdown with architecture guidance and a shortlist of relevant repositories. **Input** | Field | Type | Required | Description | |---------|--------|----------|------------------------------------------| | message | string | yes | Description of what you want to build | **Request** ```json { "jsonrpc": "2.0", "id": 4, "method": "tools/call", "params": { "name": "plan", "arguments": { "message": "Build a distributed task queue in Go with priority lanes" } } } ``` **Response** ```json { "jsonrpc": "2.0", "id": 4, "result": { "content": [{ "type": "text", "text": "{\"plan_md\":\"## Architecture\\n\\n1. Use `asynq` for Redis-backed queue...\\n\\n## Recommended repos\\n...\",\"repos\":[{\"full_name\":\"hibiken/asynq\",\"html_url\":\"https://github.com/hibiken/asynq\"}]}" }] } } ``` --- ### get-trending Return the current trending slice of the indexed corpus (mirrors GitHub trending, refreshed every 24 hours). No input required. **Input** | Field | Type | Required | Description | |-------|------|----------|-------------| | *(none)* | — | — | — | **Request** ```json { "jsonrpc": "2.0", "id": 5, "method": "tools/call", "params": { "name": "get-trending", "arguments": {} } } ``` **Response** ```json { "jsonrpc": "2.0", "id": 5, "result": { "content": [{ "type": "text", "text": "[{\"full_name\":\"anthropics/anthropic-sdk-python\",\"stars\":4200,\"language\":\"Python\",\"description\":\"The official Python library for the Anthropic API.\"}]" }] } } ``` --- ## HTTP Endpoints These endpoints are accessible without an MCP client. They require a Bearer token in the `Authorization` header except where noted. ### POST /api/contextllm/ask Deep research over the corpus. Same pipeline as the `ask` MCP tool. **Headers** ``` Authorization: Bearer cllm_YOUR_TOKEN Content-Type: application/json ``` **Body** ```json { "message": "How does React reconcile the virtual DOM?" } ``` **Response 200** ```json { "answer_md": "React uses a fiber-based reconciliation algorithm ...", "citations": [ { "full_name": "facebook/react", "html_url": "https://github.com/facebook/react" } ] } ``` **Error responses** | Status | Meaning | |--------|-----------------------------------------| | 401 | Missing or invalid Bearer token | | 429 | Rate limit exceeded | | 503 | LLM provider unavailable | --- ### POST /api/contextllm/plan Generate a coding plan + repo picks. Same pipeline as the `plan` MCP tool. **Headers** ``` Authorization: Bearer cllm_YOUR_TOKEN Content-Type: application/json ``` **Body** ```json { "message": "Build a Rust CLI that streams JSON logs to Elasticsearch" } ``` **Response 200** ```json { "plan_md": "## Approach\n\nUse `serde_json` for parsing ...", "repos": [ { "full_name": "serde-rs/serde", "html_url": "https://github.com/serde-rs/serde" } ] } ``` --- ### GET /api/contextllm/search Semantic repository search over the indexed corpus. No auth required. **Query parameters** | Parameter | Type | Required | Default | Description | |-----------|---------|----------|---------|---------------------------------------| | q | string | yes | — | Natural-language search query | | limit | integer | no | 10 | Max results to return (1–50) | **Example** ``` GET /api/contextllm/search?q=async+runtime+rust&limit=5 ``` **Response 200** ```json { "repos": [ { "full_name": "tokio-rs/tokio", "stars": 27000, "language": "Rust", "description": "A runtime for writing reliable async Rust applications.", "score": 0.94 } ] } ``` --- ## Public Pages | Path | Description | |----------------------------|------------------------------------------------------------| | `GET /contextllm` | Product page: Ask chatbox, setup & API, submit a source | | `GET /contextllm/docs` | This documentation page (interactive HTML) | | `GET /contextllm/docs.md` | Raw Markdown version of this file | *(The corpus directory is an internal admin view, not a public page.)* --- ## Rate Limits | Surface | Limit | |--------------------------|-------------------------------| | Public chatbox (IP) | 20 requests / minute | | Public ask/plan (IP) | 5 requests / minute | | Submit a source (IP) | 6 requests / minute | | API (token) | 60 requests / minute | | Embedding / ingest | Governed by provider quotas | Responses exceeding the limit return HTTP 429 with a `Retry-After` header. --- ## Changelog See `CHANGELOG.md` for version history and breaking changes.