Connect Your IDE

Any agent that speaks MCP over stdio can consume the decision graph. trurlic install writes the right config for your tool; the manual block works everywhere else.

Install

trurlic install --ide <target>

Add --dry-run to print the snippet without writing, or --binary-path <path> to embed an explicit path to the trurlic binary (default: the running executable). Existing configs are merged, not overwritten.

--ide value	Tool	Config written
claude-code	Claude Code	runs claude mcp add — see Claude Code
claude	Claude Desktop	platform claude_desktop_config.json
cursor	Cursor	~/.cursor/mcp.json
windsurf	Windsurf	~/.codeium/windsurf/mcp_config.json
cline	Cline (VS Code)	platform cline_mcp_settings.json
copilot	GitHub Copilot	platform Code/User/mcp.json — servers key
codex	Codex CLI	~/.codex/config.toml (TOML)
open-code	OpenCode	~/.config/opencode/opencode.json
open-claw	OpenClaw	~/.openclaw/workspace/config/mcporter.json
hermes-agent	Hermes Agent	~/.hermes/config.yaml (YAML)
antigravity	Antigravity CLI	~/.gemini/config/mcp_config.json

Most targets use the standard mcpServers JSON shape. The exceptions: Copilot uses a servers key, Codex uses TOML, Hermes uses YAML, and OpenCode uses its own mcp block — trurlic install handles each automatically.

Manual config

If your tool isn’t listed, add the standard MCP block to its config:

{
  "mcpServers": {
    "trurlic": {
      "command": "trurlic",
      "args": ["serve"]
    }
  }
}

No API key is needed — the server ships context and prompts; your agent’s own model does the reasoning.

Tell the agent how to use it

Add this to your agent’s instructions file (CLAUDE.md, .cursorrules, .clinerules, AGENTS.md, or equivalent):

## Trurlic

This project uses Trurlic for architectural decisions.

Before implementing any task:
1. Call `advance` with the component name and a mode (agent | interactive).
2. Follow the returned action exactly.
3. Call `advance` again after completing each action.
4. Repeat until `ready: true`, then implement — constrained by every
   decision in the context brief.
5. Call `verify_against_decisions` before committing; fix any violation.

Verify

1. Restart or reconnect the agent so it picks up the new MCP server.
2. Ask it to design or implement a component — it should call advance, then get_context.
3. Run trurlic status to confirm decisions were recorded.

For a full walkthrough see Claude Code; to build your own integration see Custom Agent.