| name | ijfw-summarize |
| description | Generate optimized project context from codebase scan. Trigger: new project, no CLAUDE.md, /ijfw-summarize |
| context | fork |
| model | haiku |
| effort | low |
Scan the codebase and generate an optimized project context file.
Process
- Read: package.json, tsconfig.json, Cargo.toml, pyproject.toml, go.mod, Dockerfile,
docker-compose.yml, .env.example, Makefile -- whatever exists.
- Scan: directory structure (2 levels deep), test framework, linter config, CI config.
- Detect: language, framework, database, ORM, auth approach, deployment target.
- Identify: key directories, entry points, API route patterns, shared utilities.
Output
Write a CLAUDE.md (or platform equivalent) with:
# Project Context
Stack: <framework> / <language> / <database>
Architecture: <pattern -- monolith, microservices, serverless, etc.>
Entry: <main entry point(s)>
Tests: <framework + command to run>
Lint: <tool + command>
## Structure
<key directories and their purpose, 1 line each>
## Patterns
<established code patterns to follow, 1 line each>
## Key Files
<important files a new contributor should know about>
Rules:
- Max 50 lines. This loads every session.
- No boilerplate explanations. Just facts.
- If uncertain about a pattern, omit it -- don't guess.