ContextForge
Turn any repository into an AI-ready context pack — AGENTS.md, .cursorrules, CLAUDE.md.
Paste a file tree and ContextForge writes the instruction files every coding agent reads: AGENTS.md, .cursorrules, CLAUDE.md, .github/copilot-instructions.md and GEMINI.md. A deterministic extractor infers package manager, framework, test runner, CI, entry points and build/test/lint commands and tags each fact as detected or inferred; an optional model-assisted pass turns those facts into prose. If the model is unavailable it degrades to a labelled rule-based draft and never fakes AI output. A Context Score (0-100) plus an open-questions list shows exactly what the input could not determine.
What it does
- 5 agent formats: AGENTS.md, .cursorrules, CLAUDE.md, copilot-instructions.md, GEMINI.md
- Deterministic extraction: package manager, framework, test runner, CI, entry points
- Context Score 0-100 with an explicit open-questions list
- Fails closed: 503 without a model key, 502 upstream, 429 at the fair-use limit
- Nothing you paste is stored
Part of the LX one-person micro-SaaS factory — a curated set of tight-vertical AI tools.