CtxKit - Server-hosted Tools for AI Agents

Documentation

Everything you need to build with CtxKit.

Quick Start

Create a Project

Create API Key

Enable url-ingest service and create an API key.

Connect Agent

Add MCP config to Claude or Cursor. Or call REST API.

MCP Integration

Add the following to your Agent's MCP configuration (e.g. claude_desktop_config.json or .cursor/mcp.json):

json

{
  "mcpServers": {
    "ctxkit": {
      "url": "https://api.ctxkit.cn/mcp",
      "headers": {
        "X-API-Key": "YOUR_API_KEY_HERE"
      }
    }
  }
}

Replace YOUR_API_KEY_HERE with your actual API key. The MCP server exposes 6 tools that your agent can call directly.

REST API

Authentication

All requests require an X-API-Key header. Keys are project-scoped and can be created in your project settings.

Ingest a URL

bash

curl -X POST https://api.ctxkit.cn/api/v1/projects/{project_id}/url-ingest/ingest \
  -H "X-API-Key: YOUR_API_KEY_HERE" \
  -H "Content-Type: application/json" \
  -d '{"url": "https://example.com/document.pdf"}'

Poll Job Status

bash

curl https://api.ctxkit.cn/api/v1/projects/{project_id}/url-ingest/jobs/{job_id} \
  -H "X-API-Key: YOUR_API_KEY_HERE"

Semantic Search

bash

curl -X POST https://api.ctxkit.cn/api/v1/projects/{project_id}/url-ingest/search \
  -H "X-API-Key: YOUR_API_KEY_HERE" \
  -H "Content-Type: application/json" \
  -d '{"query": "your search query", "limit": 10}'

Available Tools

ctxkit_ingest_url

Submit a URL for background processing. Returns a job_id for polling.

Parameters: url (required), options (optional dict)

ctxkit_ingest_get_job

Check the status of a processing job.

Parameters: job_id (required)

ctxkit_ingest_get_resource

Get metadata for a processed resource: title, summary, TOC, page count.

Parameters: resource_id (required)

ctxkit_ingest_list_resources

List all resources in the project with pagination.

Parameters: offset (default: 0), limit (default: 20)

ctxkit_ingest_read

Progressive disclosure reading. Navigate documents hierarchically.

Parameters: resource_id (required), section (optional), offset (default: 0), limit (default: 4000)

ctxkit_ingest_search

Semantic search across resources using vector similarity.

Parameters: query (required), resource_id (optional), limit (default: 10)

Ready to get started?

Get Started