High-performance GitHub repository caching service built with Cloudflare Workers and Durable Objects. Provides instant file access, full-text search, and REST API for GitHub repositories.

Add to your CLAUDE.md or AGENTS.md

to read files in GitHub repos use https://gitchamber.com. It's a website that let you list, read and search files in public github repos.

To see how to use gitchamber ALWAYS do `curl -s https://gitchamber.com` first.

• Instant file access via repository caching
• Full-text search using SQLite FTS
• Configurable TTL with automatic cache refresh
• Global edge deployment via Cloudflare Workers
• Per-repository isolation using Durable Objects
• Data is replicated in each continent. It's fast in every region

Two-tier system:

1. Worker router handles request routing
2. Durable Object (RepoCache) manages per-repository caching and storage

When accessed, repositories are downloaded as tar.gz archives from GitHub, extracted, and stored in SQLite with FTS indexing. Cache refreshes based on configurable TTL (default: 6 hours). Data is automatically cleaned up after 24 hours of inactivity.

GET https://gitchamber.com/repos/:owner/:repo/:branch/files

Returns JSON array of all file paths.

GET https://gitchamber.com/repos/:owner/:repo/:branch/files/*filepath[?showLineNumbers=true&start=N&end=M]

Returns file content. Optional parameters:

• showLineNumbers: Add line numbers
• start: Start line number
• end: End line number (defaults to start+49 if only start provided)

GET https://gitchamber.com/repos/:owner/:repo/:branch/search/*query

Full-text search returning markdown-formatted results with file paths, snippets, and line numbers.

• https://gitchamber.com/repos/remorses/gitchamber/main/files
• https://gitchamber.com/repos/remorses/gitchamber/main/files/package.json?start=5&end=50&showLineNumbers=true
• https://gitchamber.com/repos/remorses/gitchamber/main/search/cloudflare

Prerequisites: Node.js 18+, pnpm, Wrangler CLI

git clone https://github.com/your-username/gitchamber.git
cd gitchamber
pnpm install
pnpm run deployment

Environment variables in wrangler.jsonc:

{
  "vars": {
    "GITHUB_TOKEN": "", // Optional: 5K req/h limit vs 60 req/h
    "CACHE_TTL_MS": "21600000", // 6 hours default
  },
}

GitChamber offers several key advantages over the traditional WebFetch tool for accessing GitHub repositories:

• Always Fresh Content: Never get stale information - GitChamber automatically refreshes cached content based on configurable TTL, ensuring you always see the latest version of repository files
• Native Markdown Support: Files are already in markdown format, eliminating the need to convert HTML back to markdown, preserving original formatting and structure
• Access to Implementation Files: Using custom glob patterns, you can retrieve actual source code files (.js, .ts, .py, etc.), not just documentation
• Context-Aware Pagination: Supports line-based pagination with start and end parameters to prevent LLM context overflow - Claude and other LLMs can efficiently process large files in chunks
• Precise Search: Uses exact string matching instead of imprecise fuzzy or semantic search, ensuring you find exactly what you're looking for with full-text search capabilities via SQLite FTS
• Optimized for LLMs: Purpose-built for AI agents and LLMs like Claude that need structured, reliable access to repository contents

• Cloudflare Workers (runtime)

• Durable Objects with SQLite (storage)

• TypeScript

• Spiceflow (API framework)

• Zod (validation)

• Cold start: 200-500ms
• Cached response: 10-50ms globally
• Automatic cleanup after 24h inactivity

The API follows OpenAPI 3.0 specification. Key endpoints:

openapi: 3.0.0
info:
  title: GitChamber API
  version: 1.0.0
servers:
  - url: https://gitchamber.com
paths:
  /repos/{owner}/{repo}/{branch}/files:
    get:
      summary: List repository files
      responses:
        "200":
          content:
            application/json:
              schema:
                type: array
                items:
                  type: string
  /repos/{owner}/{repo}/{branch}/files/{filepath}:
    get:
      summary: Get file content
      parameters:
        - name: showLineNumbers
          in: query
          schema:
            type: boolean
        - name: start
          in: query
          schema:
            type: integer
        - name: end
          in: query
          schema:
            type: integer
      responses:
        "200":
          content:
            text/plain:
              schema:
                type: string
  /repos/{owner}/{repo}/{branch}/search/{query}:
    get:
      summary: Search repository content
      responses:
        "200":
          content:
            text/markdown:
              schema:
                type: string