The MCP Bridge: Connecting LLMs to Your Active Project Command Hub
Unlock the Model Context Protocol (MCP). Use our native MCP server to connect external agents to your workspace via JSON-RPC 2.0.
- The Open Protocol Standard: The Model Context Protocol (MCP) is the universal open standard connecting AI models to external data sources and tools.
- Native CoMng.AI MCP Server: Provides an enterprise-grade endpoint at `https://comng.ai/app/api/external/mcp.php` utilizing JSON-RPC 2.0.
- Discoverable Toolsets: Exposes governed project tools including task fetching, status updates, RAID logging, and pulse telemetry queries.
- Hardened Security: Protects project integrity through Bearer token authentication, scoped project permissions, and strict IP whitelisting.
Bridging Strategy and Code via the Model Context Protocol
As autonomous AI agents become standard across software engineering, the bottleneck is no longer coding speed—it is project context. An AI agent running in a developer's terminal doesn't know the overarching milestone dependencies, budget constraints, or contractual acceptance criteria.
CoMng.AI solves this by deploying a native Model Context Protocol (MCP) server. External agents connect directly to your project command hub over JSON-RPC 2.0, turning your project plan into discoverable, actionable tools.
Technical Architecture of the CoMng.AI MCP Implementation
| Specification | Detail | Enterprise Purpose |
|---|---|---|
| Endpoint URI | https://comng.ai/app/api/external/mcp.php | Dedicated high-availability JSON-RPC 2.0 gateway |
| Transport Protocol | Streamable HTTP / JSON-RPC 2.0 | Low-latency bidirectional tool communication |
| Authentication | X-API-Key or Bearer Token Header | Cryptographically signed per-project credentials |
| Access Scope | Project-Scoped with Read/Write Toggles | Enforces strict isolation between organization projects |
| Network Guardrails | CIDR IP Whitelisting & Sandboxing | Prevents unauthorized connections from rogue IP ranges |
Core MCP Tools Exposed to AI Coding Agents
When Claude Code or Gemini connects to CoMng.AI via MCP, it gains access to governed tools:
- `get_project_pulse()`: Returns the 5-dimension health telemetry to verify whether ongoing work threatens schedule or budget. - `get_assigned_tasks(agent_id)`: Fetches active WBS items, dependencies, and contractual acceptance criteria. - `update_task_progress(task_id, percent, notes)`: Allows the agent to report progress and link Git commit hashes directly to the Gantt chart. - `flag_risk_blocker(task_id, description)`: Instantly logs unexpected technical blockers into the project's RAID+ register.
Transform Your Project Execution Today
Experience how CoMng.AI unifies humans and AI agents with 5-Dimension Telemetry, zero-drift migrations, and load-bearing intelligence.
Frequently Asked Questions
What does the MCP server do?
It translates CoMng.AI's backend REST capabilities into standard, discoverable MCP tools that any compliant AI agent can inspect and execute.
Is the connection secure?
Yes. Connections require dedicated API keys, are scoped strictly to individual projects, and can be restricted to specific developer IP addresses.
Can I use any LLM with this?
Yes. Any agent or LLM client that supports the Model Context Protocol standard (including Claude Code, Gemini CLI, Cursor, and custom frameworks) can connect.