AI tool integration
Wire SEIF into Claude, Cursor, Copilot, Gemini, and Codex via standard configuration files.
SEIF integrates with AI tools through the configuration files they already read. Place the relevant file at your project root and the tool loads SEIF context and provenance hooks automatically.
Claude / Claude Code
Reads CLAUDE.md. SEIF hooks into pre/post tool-use events for automatic signing and quality measurement.
Cursor
Reads .cursor/rules/*.mdc and an MCP server entry. Run seif setup cursor to generate both.
GitHub Copilot
Reads .github/copilot-instructions.md. SEIF modules are injected via instruction prepend.
Gemini CLI
Reads GEMINI.md at the project root for context and signing hooks.
OpenAI Codex
Reads AGENTS.md for agent directives, including SEIF protocol instructions.
Model Context Protocol (MCP)
The CLI ships an MCP server so any MCP-capable client can call SEIF governance tools directly.
seif mcp serveFor Cursor specifically:
seif setup cursor
# writes .cursor/mcp.json and .cursor/rules/*.mdcAfter setup, restart your client and enable the SEIF server. The first tool each
session is session_orient, which returns the grounding kernel and a snapshot of
your workspace memory.
What gets loaded
On session start, a SEIF-aware tool loads:
- the conduct constitution (operating maxims, with their control-theory origin model),
- your memory modules (decisions, conventions, observations),
- open governance cycles and high-priority pending items.
This is how a fresh session on any machine starts already knowing your project.
Intent → verb map
The compact, measured capability map an AI reaches for at the moment of need — plain intent on the left, the exact SEIF verb on the right, plus the multi-step workflow recipes.
MCP setup
Run the SEIF MCP server from the seif binary — no npm install — and wire it into Cursor or Claude Desktop.