settings — independently scanned and version-tracked by SaferSkills.
SaferSkills independently audited settings (Hook) and scored it 100/100 (green). The audit ran 55 deterministic rules across Security, Supply Chain, Maintenance, Transparency, and Community; it found 0 high-severity and 0 lower-severity findings. The full rule-by-rule trace and per-finding evidence are below. Free, methodology-open.
Findings & checks · 0 flagged
Every scanned point with the score it earned and what moved between them.
First recorded scan — no prior version to compare against.
The primary manifest — the file an agent reads to learn what this artifact does.
Your AI gets smarter every session — automatically.
AI Rule Learning watches how your AI conversations go, spots where things go wrong, and turns those moments into rules that prevent the same mistakes from happening again. It also remembers facts about you and saves reusable workflows — so every future session starts with full context.
Works with Claude Code, Cursor, Windsurf, GitHub Copilot, Codex, and any MCP-compatible agent. No manual prompting. No configuration tweaking.
Free for individuals. Commercial and government use requires written permission.
Every session, AI Rule Learning builds on five compounding capabilities:
| Pillar | What it does |
|---|---|
| Rules | Detects friction patterns (corrections, repeated context, incomplete responses) and writes guardrail rules to every AI agent config |
| Memory | Remembers facts about you — preferences, stack, projects, constraints — and loads them at the start of every session |
| Universal injection | Writes rules and memory into Claude Code, Cursor, Windsurf, and GitHub Copilot configs automatically |
| Auto-sync | Runs nightly via LaunchAgent / systemd / cron — no manual action needed |
| Skills | Saves repeatable multi-step workflows (e.g. "deploy to HF", "create FastAPI endpoint") so any agent can recall and follow them exactly |
pip install ai-rule-learning-mcpClaude Code — add to ~/.claude/claude_desktop_config.json or claude_code_config.json:
{
"mcpServers": {
"ai-rule-learning": {
"command": "ai-rule-learning-mcp"
}
}
}Cursor / Windsurf — add the same block to your MCP settings file.
Zero config option — use the CLI directly without any MCP setup (see below).
Once connected, your AI agent has access to these tools automatically:
| Tool | When to call it |
|---|---|
get_guardrail_rules | Start of a session — loads your personalised rules |
record_feedback | When you correct the agent or repeat context — creates a rule in real time |
sync_sessions | Scans session history, detects patterns, writes rules to all agent configs |
remember | Stores a fact, preference, or constraint about you |
recall | Loads everything known about you at session start |
save_skill | Saves a multi-step workflow for future reuse |
list_skills | Shows all saved skills (names + descriptions) |
get_skill | Loads the full steps for a saved skill |
install_scheduler | Installs nightly auto-sync (macOS/Linux) |
list_providers | Shows which agents and session paths are detected |
preference coding/style/format preference
project active project or task context
never hard constraint — something to never do
user_info fact about you (timezone, role, team size)
context stack, tools, environment info# Scan sessions and generate rules
ai-rule-learning sync
# Show current rules and detected agent configs
ai-rule-learning status
# Print active rules
ai-rule-learning rules
# Remove all injected sections
ai-rule-learning clear
# Memory
ai-rule-learning memory show
ai-rule-learning memory add preference "Always use type hints in Python"
ai-rule-learning memory clear
# Skills
ai-rule-learning skills
ai-rule-learning skills show "Deploy to HuggingFace"
ai-rule-learning skills delete "Old Workflow"
# Auto-sync scheduler
ai-rule-learning install-cron
ai-rule-learning cron-status
ai-rule-learning uninstall-cronRules, memory, and skills are injected as fenced blocks (using HTML comment markers) into every detected AI agent config:
| Agent | Config file |
|---|---|
| Claude Code | ~/.claude/CLAUDE.md |
| Cursor | ~/.cursor/rules/ai-guardrails.md |
| Windsurf | ~/.windsurf/rules/ai-guardrails.md |
| GitHub Copilot | ~/.config/github-copilot/instructions.md |
All sections are idempotent — re-running sync updates the block without duplicating it.
All data is stored at ~/.ai-rule-learning/:
~/.ai-rule-learning/
rules.jsonl Active guardrail rules
memory.jsonl Remembered facts and preferences
conversations.jsonl Parsed session history
processed.jsonl Deduplication index
skills/ Saved skill procedures (one .md per skill)
sync.log Auto-sync outputNo data leaves your machine unless you explicitly opt in to community contribution (see below).
Set environment variables to back up rules to a private HF dataset:
export HF_TOKEN=hf_...
export ARL_DATASET=yourname/my-rulesOpt in to share anonymised gap pattern counts (no conversation text, no code) with the community pool. This improves rule templates for everyone.
ai-rule-learning sync --contributeOr in MCP: sync_sessions(contribute=True)
See CONTRIBUTING.md.
~30 seconds. Free. No account. Every finding cites a rule and a line of evidence.