Repowise

Repowise turns a repository into structured intelligence: a defect-validated 1-10 health score per file (explicitly zero LLM in that path — it's derived from real evidence), a 0-10 change-risk score for any commit or PR, dependency and call graphs, code ownership, and documentation with source links. It also answers "how much of this codebase did AI write, and is that part healthy?" — provenance scoring that's about to matter a lot more as AI-generated code compounds.

For agents, it ships ten MCP tools that hand Cursor or Claude Code grounded codebase context instead of the usual guess-and-grep.

What makes it worth a look is the "shows its work" stance: scores link back to the underlying evidence rather than presenting a vibe. Who it's for: tech leads inheriting unfamiliar codebases, teams tracking whether AI-generated code is rotting their health scores, and platform engineers who want their coding agents grounded in facts. Starts at $14/month. As with any scoring tool, use it to direct attention, not to replace reading code that matters.