SSK0188

Agent Rule Sentinel — AI skill for Claude Code

AI community

A local-first open-source tool for safely synchronizing and validating AI coding agent instructions across Codex and Claude Code.

How to install Agent Rule Sentinel

This entry records only its repository, not the path inside it, so there is no exact command to give. Open SSK0188/agent-rule-sentinel and copy the folder into ~/.claude/skills/, or the file into ~/.claude/agents/.

What Agent Rule Sentinel does

A local-first open-source tool for safely synchronizing and validating AI coding agent instructions across Codex and Claude Code.

Alternatives in AI

Chat Ollama — ChatOllama is an open source agentic app for running AI agents across local and hosted models 3.5k ★
Token Monitor — Local-first desktop widget for tracking token usage, costs, and limits across 32+ AI coding tools—including Cl 1.8k ★
Open Science — Open Science Desktop — local-first, model-agnostic AI research workbench for macOS, Windows & Linux 1.5k ★

README

Agent Rule Sentinel

**A local-first, review-before-write synchronizer for Codex and Claude Code project instructions.**

[中文说明](docs/README.zh-CN.md) · [Contributing](CONTRIBUTING.md) · [Security](SECURITY.md) · [Roadmap](ROADMAP.md)

Keep a canonical set of project instructions in `.agent-rule-sentinel/source.md`, then produce consistent managed blocks in `AGENTS.md` (Codex) and `CLAUDE.md` (Claude Code). **You always preview changes first.** No provider account, API key or network connection is required.

Why?

Developers working across multiple coding agents often duplicate project rules. Edits drift, and an automated overwrite could erase personal or agent-specific instructions. Agent Rule Sentinel updates only a clearly marked section, leaving surrounding content untouched.

Features

Zero runtime dependencies (Python 3.10+), offline by design.
plan previews unified diffs, with no writes.
sync --apply only changes explicitly selected target files.
Existing custom text outside managed markers is preserved.
Existing targets are privately backed up before replacement; newly created targets are not backed up.
check detects drift for CI, returning status 1 when a target needs synchronization.
Read-only doctor, safe import (adopt), conflict-aware restore (undo) in the development branch.
Opt-in Gemini CLI (GEMINI.md) project instruction support.
Refuses symlink targets, malformed markers, oversized files and several recognizable credential patterns.
No background watcher or automatic commits, push, network calls, or AI model execution.

**Version note:** The commands below marked 0.2.0.dev0 exist on the unreleased `main` development branch, **not** in the published `v0.1.0` tag.

**Security scope:** The credential detector is heuristic, not an exhaustive secret scanner. Review the source and diffs before committing generated files. Backups may contain sensitive content and are ignored by Git.