MCP Server
Model Context Protocol integration for AI agent workflows.
Overview
Trellis includes a built-in MCP (Model Context Protocol) server that exposes all platform capabilities as tools for AI agents. This enables seamless integration with AI coding assistants and automation pipelines.
Starting the Server
trellis mcp
Available Tools
| Tool | Purpose |
|---|---|
trellis_status | Branch, op count, tracked files, recent ops |
trellis_log | Op history with optional filters |
trellis_files | List all tracked files |
trellis_branch | List / create / switch / delete branches |
trellis_milestone | List or create milestones |
trellis_diff | File-level diff between op hashes |
trellis_garden | List / search / stats / revive idea clusters |
trellis_parse | Parse TS/JS content into AST entities |
trellis_semantic_diff | Semantic diff between file versions |
trellis_init | Initialize a new repository |
trellis_issue_create | Create issue with metadata and criteria |
trellis_issue_list | List/filter issues |
trellis_issue_start | Start issue (auto-branch, auto-assign) |
trellis_issue_pause | Pause issue |
trellis_issue_resume | Resume issue |
trellis_issue_triage | Move issue backlog → queue |
trellis_issue_update | Update issue metadata |
trellis_issue_check | Run acceptance criteria |
trellis_issue_close | Close issue |
trellis_decision_list | List/filter decision traces |
trellis_decision_show | Show full decision trace |
trellis_decision_chain | Trace decisions affecting an entity |
Configuration
Add to your .mcp.json:
{
"mcpServers": {
"trellis": {
"command": "trellis",
"args": ["mcp"],
"cwd": "/path/to/your/project"
}
}
}