Skip to content

Releases: ichbrt/CodeRecall

Release list

CodeRecall v0.1.0-alpha.1 - Don't regenerate. Recall and reuse.

CodeRecall v0.1.0-alpha.1 - Don't regenerate. Recall and reuse. Pre-release
Pre-release

Choose a tag to compare

View all tags
@ichbrt ichbrt released this 13 Sep 07:59
v0.1.0-alpha.1
1ccf920

CodeRecall v0.1.0-alpha.1

Don't regenerate. Recall and reuse.

CodeRecall gives coding agents local engineering memory. It inspects repositories you've already built, explains which ones fit a new task, copies suitable code directly, and produces a plan so the agent can implement only the difference.

This first public alpha proves the complete local loop: index → fingerprint → match → plan → copy → adapt → verify → record provenance. It is intended for early adopters and contributors working with repositories they own or are permitted to reuse.

Try it

Requires Node.js 24+ and Git. The attached package contains the compiled CLI:

npm install --global https://github.com/ichbrt/CodeRecall/releases/download/v0.1.0-alpha.1/coderecall-0.1.0-alpha.1.tgz
coderecall --help
coderecall doctor

For the reproducible three-repository demo, clone the source and follow the quick start. Connect your coding agent with the MCP setup guide.

What's included

  • Local SQLite memory and JavaScript/TypeScript static analysis.
  • Explainable scores and independent safety/compatibility gates.
  • Machine-readable adaptation plans, safe copy previews and exact file transfers.
  • Verification and file-level provenance.
  • CLI and six MCP tools for compatible coding agents.
  • MIT-licensed source, tests, synthetic fixtures and documentation.

What to expect

The demo selects a Next.js SaaS base, identifies Express booking code as requiring module review, and rejects an unrelated utility. It copies existing files before a small, fixed demo patch adds booking behavior. This demonstrates correct reuse mechanics, not a production SaaS or an AI performance benchmark.

Automatic module assembly, general semantic migrations and measured token/cost savings remain future work. Verification runs trusted code on your machine; review the security model before indexing or executing unfamiliar repositories.

Useful feedback: install problems, reproducible compatibility mistakes, ignore-rule edge cases and small synthetic examples of code that should be reused. Please keep private code and credentials out of public issues.

Loading