Corbell-AI

Corbell-AI / Corbell

Public

AI-powered spec generation and review using multi-repo code graph intelligence for backend teams that ship to production.

code-analysis code-embedding code-intelligence-mcp knowledge-graph mcp-server
18
1
100% credibility
Found Mar 15, 2026 at 18 stars -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
Python
AI Summary

Corbell is a tool for generating multi-repo architecture graphs, AI-powered spec generation, and architecture reviews for backend teams by analyzing code repositories and design documents.

Star Growth

See how this repo grew from 18 to 18 stars Sign Up Free
Repurpose This Repo

Repurpose is a Pro feature

Generate ready-to-use prompts for X threads, LinkedIn posts, blog posts, YouTube scripts, and more -- with full repo context baked in.

Unlock Repurpose
AI-Generated Review

What is Corbell?

Corbell scans your multi-repo codebase to build a local knowledge graph of services, dependencies, and call paths, then uses AI like Claude or GPT-4o to generate technical specs that respect your existing patterns. Feed it a PRD, and it auto-discovers relevant services via semantic code search, injects real graph context and code snippets, and outputs a full design doc with token costs tracked. Python CLI tool, runs offline on SQLite—no cloud setup needed.

Why is it gaining traction?

Unlike generic AI chatbots, it grounds specs in your actual architecture graph, avoiding hallucinated designs, and integrates with Linear/Notion exports plus a local D3 graph browser. MCP server hooks into Cursor/Claude Desktop for AI-powered IDE context on call paths and constraints. Auto service discovery from PRDs saves manual service hunting, making it a practical AI boost for GitHub monorepos.

Who should use this?

Staff engineers or architects at backend teams juggling 5-10 repos, tired of rehashing Kafka vs SQS debates from lost Confluence pages. Ideal for quarterly planning where new specs must align with git-coupled flows and constraints like "p99 < 200ms." Production shippers needing AI-powered specs before coding sprints.

Verdict

Promising local AI for architecture docs, but 18 stars and 1.0% credibility signal early alpha—docs are solid, but expect rough edges without tests. Try for small workspaces if you're evaluating AI-powered GitHub spec tools.

(187 words)

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.