Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx agentmods add instructions/cololi/ruleforge/claude-mdgit clone --depth 1 https://github.com/cololi/ruleforgeWhat it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5 | $0.00352 | $0.00352 |
| Opus 5 | $0.00176 | $0.00176 |
| Sonnet 5 | $0.00070 | $0.00070 |
| Haiku 4.5 | $0.00035 | $0.00035 |
Grade A, and why
ruleforge CLAUDE.md scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 2d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
What it actually says
RuleForge
CLI tool that reads Claude Code / Codex session logs and generates optimized rules files.
Tech Stack
- Python 3.12+, uv for package management
- Typer (CLI), Rich (terminal UI), Pydantic (config + models)
- anthropic SDK for LLM classification
- SQLite for effectiveness tracking
Commands
- Build:
uv build - Test:
uv run pytest - Lint:
uv run ruff check src/ tests/ - Type check:
uv run mypy src/ - Install dev:
uv sync --dev
Code Style
- Use functional style, avoid classes unless modeling state (Tracker, Config)
- Use
from __future__ import annotationsin every file - Prefer
pathlib.Pathoveros.path - Use Pydantic BaseModel for all data structures
- Async functions for any LLM API calls
- Type all function signatures strictly
Architecture
parsers/— Read JSONL session logs into Session modelsextractors/— Heuristic + LLM pattern extraction pipelinegenerators/— Render patterns into CLAUDE.md / rules filestracker/— SQLite-backed effectiveness measurementhooks/— Claude Code hook installermodels/— Pydantic domain models (Session, Pattern, Rule)
Prohibitions
- NEVER use print() for user output — use Rich console
- NEVER hardcode API keys or model names — use config
- Do not add dependencies without updating pyproject.toml
- Do not write tests that require live API calls — mock LLM responses
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 2d ago First seen · 39 lines · 352 tokens per session scan A ee33e2e1be11
ruleforge CLAUDE.md is an instructions file published in the GitHub repository cololi/ruleforge (2 stars, last pushed 4mo ago), licensed MIT. It adds 352 tokens to every session, about $0.0018 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.
Other instructions, from other repositories
ashfox AGENTS.md
Instructions for sigee-min/ashfox, a project described as: Powerful Blockbench MCP server and AI-native low-poly workbench for modeling, pixel texturing, rigging, animation, and game-ready exports from a prompt.
vision-memory-mcp copilot-instructions.md
Copilot instructions for putervision/vision-memory-mcp, covering visual memory (vision-memory-mcp), 1. mandatory workflow & priority, 2. tool reference summary (15 core mcp tools), state memory (state-memory-mcp) and 1. priority order.
vision-memory-mcp CLAUDE.md
Claude Code instructions for putervision/vision-memory-mcp, covering visual memory (vision-memory-mcp), 1. mandatory workflow & priority, 2. tool reference summary (15 core mcp tools), state memory (state-memory-mcp) and 1. priority order.
workflow-atlas AGENTS.md
AGENTS.md instructions for claudiu-deving/workflow-atlas, covering agent setup, run it, tests, what this is for and the workflow model: an infinite-nested canvas.
commander AGENTS.md
AGENTS.md instructions for autohandai/commander, covering all agents development standards, 🎯 mandatory requirements for all agents, you write the tdd and then write the feature implementation, architecture pattern - strict compliance and modular structure (required).
memory-bank-mcp copilot-instructions.md
Copilot instructions for diaz3618/memory-bank-mcp, covering memory bank — copilot instructions, mandatory workflow (every task, no exceptions), start of task, during task and end of task.