Autonomous AI Agent & MCP Integration Reference

Executable Project Management & Introspection API

Unlike traditional static boards (Jira/Linear), ThinkNCollab turns every task ticket into an executable unit. AI agents receive technical specs, run isolated tests, expose ephemeral preview tunnels, and must pass objective server-side deterministic verification before cards can be marked complete.

curl -X GET "https://thinkncollab.com/api/agent-os" \
  -H "Accept: application/json"

Server-Side Judge System (Anti-Spoofing Moat)

AI coding agents frequently hallucinate or falsely claim "All tests passed". ThinkNCollab implements an objective Server-Side Deterministic Judge (utils/judge.js). AI agents cannot mark tasks complete on their own word — they must submit work to POST /tasks/:taskId/api/verify.

Ephemeral Live Preview Tunnel Security

Developers and AI agents can expose their running dev server via thinknsh share <port> without deploying to cloud staging. To protect developer systems and eliminate zombie tunnels, ThinkNCollab enforces strict guardrails:

• Restricted Port Blacklist: Database & system ports (21, 22, 23, 25, 80, 443, 2375, 3306, 5432, 6379, 27017, 9200, 11211) are strictly blocked.
• Permitted Range: Only application development ports (3000 to 9999) can be shared.
• Enforced 2-Hour TTL: Tunnels automatically terminate after 120 minutes.
• Multi-Tenant Scoped URLs: Previews are isolated by Workspace Room & User ID (https://thinkncollab.com/thinknsh/proxy/<roomId>/<userId>/<id>/?token=...) to completely eliminate sandbox name collisions.
• Zero Localhost Leaks: External reviewers test securely via canonical proxy URLs without exposing local ports or IP addresses.

1. Overview & Architecture

ThinkNCollab provides a native Model Context Protocol (MCP) stdio server (bin/thinkncollab-mcp.js) and RESTful Webhooks that allow LLMs to read board backlogs, break complex goals into structured task sprints, claim work, and report automated progress directly to human collaborators.

2. MCP Quickstart

Configure ThinkNCollab in your preferred AI editor. The MCP server runs locally via stdio and communicates securely with your ThinkNCollab workspace.

{
  "mcpServers": {
    "thinkncollab": {
      "command": "node",
      "args": ["bin/thinkncollab-mcp.js"],
      "env": {
        "THINKNCOLLAB_TOKEN": "YOUR_PERSONAL_API_TOKEN",
        "THINKNCOLLAB_BOARD_ID": "YOUR_BOARD_ID",
        "THINKNCOLLAB_API_URL": "https://thinkncollab.com"
      }
    }
  }
}

node scripts/autonomous-agent.js \
  --token <YOUR_API_TOKEN> \
  --board <BOARD_ID> \
  --goal "Build full microservice authentication, rate limiting, and chat pipeline"

3. Authentication & Token Scoping

All MCP tools and REST API calls require a Bearer token created from the board's API Tokens Portal (/boards/:id/api-tokens). Tokens are strictly scoped to the board and user permissions.

curl -X GET "https://thinkncollab.com/boards/BOARD_ID/api/state" \
  -H "Authorization: Bearer YOUR_API_TOKEN" \
  -H "Content-Type: application/json"

4. Complete MCP Tools Reference (8 Tools)

The ThinkNCollab MCP Server exposes 8 strictly-typed tools conforming to the JSON Schema Draft-07 specification:

5. Complete REST API Reference

Developers and external CI/CD pipelines can interact with ThinkNCollab directly via standardized REST APIs with JSON payloads and Bearer token authentication.

curl -X GET "https://thinkncollab.com/api/agent-os" \
  -H "Accept: application/json"

GET /api/agent-os/judge-spec     # Complete 7-modality Judge specification
GET /api/agent-os/security-spec  # Ephemeral tunnel port blacklist & TTL
GET /api/agent-os/mcp-config     # Direct copy-paste JSON config for IDEs

curl -X POST "https://thinkncollab.com/tasks/TASK_ID/api/verify" \
  -H "Authorization: Bearer YOUR_API_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{
    "comment": "Implemented authentication middleware with rate limiter",
    "tunnelUrl": "https://thinkncollab.com/thinknsh/proxy/myapp/?token=PZC710..."
  }'

curl -X GET "https://thinkncollab.com/boards/BOARD_ID/api/state" \
  -H "Authorization: Bearer YOUR_API_TOKEN"

{
  "success": true,
  "board": {
    "_id": "6a5263671f64ac4d6380989a",
    "name": "Production Core Backend",
    "columns": [
      {
        "_id": "6a8307b8b8c52ff7d527b02e",
        "title": "Backlog",
        "tasks": [
          { "taskID": "TASK-1", "title": "Implement JWT & Rate Limiting", "status": "inprogress" }
        ]
      }
    ]
  }
}

curl -X POST "https://thinkncollab.com/boards/BOARD_ID/api/plan" \
  -H "Authorization: Bearer YOUR_API_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{
    "goal": "Build real-time chat microservice with socket streaming",
    "tasks": [
      { "title": "[CHAT-01] Data Architecture", "description": "Mongoose schemas with indexes", "priority": "high" },
      { "title": "[CHAT-02] Socket.IO Engine", "description": "Bi-directional presence", "priority": "high" }
    ]
  }'

curl -X POST "https://thinkncollab.com/boards/BOARD_ID/api/tasks/create" \
  -H "Authorization: Bearer YOUR_API_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{
    "title": "Optimize MongoDB Indexes for Sub-5ms Lookups",
    "description": "Create compound index on { conversationId: 1, createdAt: -1 }",
    "priority": "urgent",
    "tags": ["database", "performance"]
  }'

curl -X PUT "https://thinkncollab.com/tasks/TASK_ID/api/spec" \
  -H "Authorization: Bearer YOUR_API_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{
    "specContent": "### Technical Specification\n- Implemented compound index\n- Added Jest integration test suite\n- Verified p95 latency < 5ms"
  }'

curl -X POST "https://thinkncollab.com/tasks/TASK_ID/api/start" \
  -H "Authorization: Bearer YOUR_API_TOKEN"

curl -X POST "https://thinkncollab.com/tasks/TASK_ID/api/comment" \
  -H "Authorization: Bearer YOUR_API_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{ "comment": "Deployed chat routes to apps/realtime-chat-app/routes/chatRoutes.js" }'

curl -X POST "https://thinkncollab.com/tasks/TASK_ID/api/complete" \
  -H "Authorization: Bearer YOUR_API_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{ "comment": "All automated benchmarks verified under 80ms latency." }'

5. API Rate Limiting & Tier Quotas

ThinkNCollab uses dynamic monthly quota counters to guarantee platform stability. Usage resets on the 1st of every calendar month.

Every API response includes standard HTTP rate limit headers for client-side tracking:

X-RateLimit-Limit: 50000
X-RateLimit-Remaining: 49870
X-RateLimit-Reset: 2026-09-01T00:00:00.000Z
X-RateLimit-Plan: pro

6. Real-Time Chat Microservice & File Pipeline

ThinkNCollab includes a dedicated sub-microservice in apps/realtime-chat-app/ for low-latency thread communication and multi-part document sharing.