FAQ
Why not just use cat * or a simple bash script?
Section titled “Why not just use cat * or a simple bash script?”Simply concatenating files loses all structural hierarchy. An LLM sees thousands of lines of code with no indication of what is an entry point versus a test script. Furthermore, simple scripts won’t respect .gitignore rules reliably, nor can they safely compress tokens when you run into context limits.
Does codectx send my code to the cloud?
Section titled “Does codectx send my code to the cloud?”No. codectx runs entirely locally on your machine. It simply formats your files into a CONTEXT.md file. What you do with that file (e.g., send it to OpenAI, Anthropic, or an open-source local model footprint) is entirely in your control.
What languages are supported?
Section titled “What languages are supported?”Currently, due to the Tree-sitter backend, codectx officially supports:
- Python
- JavaScript / TypeScript
- Rust
- Go
Fallback parsing mechanisms allow it to ingest and format any text-based file, but the advanced Dependency Graph optimizations are limited to languages with strict parser definitions in the system.
Can I use this in CI/CD?
Section titled “Can I use this in CI/CD?”Yes! Generating the CONTEXT.md on every PR via a GitHub action provides a highly useful, deterministic snapshot file that reviewers or automated AI code reviewers can utilize effortlessly.