Packages

In-process, OTP-supervised security runtime for MCP and agent-tool boundaries with deterministic policy, Agent Trust attestations, and signed evidence.

Current section

Files

Jump to
sigil_guard AGENTS.md
Raw

AGENTS.md

# Agent Instructions
Read and follow `CLAUDE.md` first. It is the canonical repo contract for
architecture, terminology, quality gates, and the referenced `.claude/`
standards and skills.
Do not duplicate or reinterpret those rules here. This file exists so agents
outside Claude Code can discover the same project instructions quickly.
Canonical skills live only under `.claude/skills/` and are selected automatically
from their descriptions. Clients without native skill discovery should read the
matching `SKILL.md` there and apply the same validation gates before handing off.