Skip to content

Stable context-export format for downstream tools #198

@googlarz

Description

@googlarz

Context

I've been building a tool that consumes .codegraph/ output to understand project structure. The data in .codegraph/ (communities, hub nodes, entry points) is exactly what I need — but I'm reading it by guessing at file names and JSON shapes rather than relying on a documented contract.

The ask

Would you consider documenting a stable schema for .codegraph/index.json (or whichever file is meant to be the primary output)?

Specifically: the fields that are guaranteed to be present, their types, and whether the format is considered stable across versions.

Even a brief note in the README like "The .codegraph/ directory contains X. The top-level keys are Y and Z. This format is stable/experimental." would be enough for downstream tools to consume it reliably.

Why it matters

Right now anyone wanting to build on codegraph output has to reverse-engineer the schema from examples or source code. A documented, stable schema would let:

  • IDE extensions index it without re-running the graph
  • Documentation tools render architecture summaries
  • Context generators (like mine) include structural info in AI context files
  • Scripts in CI surface community/hub data without shelling out

Happy to help

If this is something you'd find useful, I'm happy to contribute:

  • A JSON Schema or TypeScript type definition for the output format
  • A brief section in the README documenting the fields
  • A versioning note (e.g. "format is stable as of v0.x")

Just let me know what would be most helpful. Thanks for building codegraph — the community detection approach is genuinely useful for large codebases.

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions