codeflow vs codegraph-mcp
Paste a GitHub URL or drop a local folder → interactive architecture map in the browser: dependency graph, blast radius, health grade, security scan. Single index.html, zero install. — versus — No-train, on-prem code knowledge graph served to AI agents over MCP — symbols, call edges, cross-language links and blast-radius queries, with a hash-chained audit log of every read.
Same 'understand the codebase structure' job: codegraph-mcp is an on-prem symbol/call-edge knowledge graph consumed by agents; codeflow is a zero-install visual map consumed by people.
| codeflow | codegraph-mcp | |
|---|---|---|
| Stars | 4.6k | 5 |
| Forks | 713 | 0 |
| Language | HTML | Python |
| License | — | NOASSERTION |
| Last activity | 6 days ago | 16 days ago |
| Topics | coding | coding, rag |
| Curated connections | 3 | 6 |
codeflow — the curator's take
The fastest 'what am I looking at?' tool for an unfamiliar repo: zero install, code never leaves the browser, and blast-radius answers 'what breaks if I touch this?' before you refactor. The JSON export and README-card GitHub Action make it more than a toy. NOT compiler-grade: dependency edges come from lightweight in-browser parsing across 30+ languages, so trust it for orientation, not for exhaustive call-graph truth — agents needing tree-sitter-accurate graphs should use code-review-graph or codegraph-mcp. Big repos also eat GitHub API rate limits fast without a token.
codegraph-mcp — the curator's take
Niche but real: if agents must understand code that cannot leave your building AND compliance asks 'what exactly did the agent read', the tamper-evident audit chain is the only game in town; cross-language call edges (TS fetch → Go/Python handler) catch what one-file context misses. NOT for most teams yet: 5 stars, single vendor, and — deal-breaker until fixed — no clear open-source license (NOASSERTION on GitHub). Treat it as an evaluation candidate, not a dependency.