Repowise · Codebase intelligence
A runtime substrate that turns an agent's execution into a reversible, Git-like trace, so meta-agents can observe, fork, replay, and revert any run. Couples agent and environments in a copy-on-write fork ~5x faster than docker commit, with ~95% KV-cache reuse on replay. Framework built for meta-agents to supervise, optimize, and train other agents
Repowise ranks every file by predicted defect risk. Of the 20 files it flags as riskiest, 4 (20%) actually needed a bug fix in the last 6 months — 11.2× better than picking files at random.
commons-vcs
commons-vcs
shepherd2
shepherd2
A runtime substrate that turns an agent's execution into a reversible, Git-like trace, so meta-agents can observe, fork, replay, and revert any run. Couples agent and environments in a copy-on-write fork ~5x faster than docker commit, with ~95% KV-cache reuse on replay. Framework built for meta-agents to supervise, optimize, and train other agents This page is an auto-generated, always-fresh map of the shepherd-agents/shepherd repository, written primarily in Python. Repowise indexes the source, parses every symbol, computes a dependency graph, scores per-file code health from complexity, duplication, test coverage and churn, mines git history for hotspots and ownership, and lifts the resulting architectural decisions into a wiki you can read or query through MCP.
The codebase has 1,841 source files, 27,182 symbols, and 10 languages, structured as a monorepo with 2 packages. Git churn analysis flags 114 high-frequency files as hotspots — places where bugs, rewrites, and code review tend to concentrate. The dependency graph clusters into 285 tightly-coupled module communities.
Use the panels above to open the interactive dashboards, or connect this repo to your editor via the Repowise MCP server for grounded answers inside Claude, Cursor, or VS Code.
Files, symbols, languages, packages, git intelligence
Dependency graph, layered architecture, symbol index, and third-party dependencies
Per-file health scores, hotspots, test coverage, dead code, and refactoring targets
Module-by-module documentation generated from source
Change-risk ranked history with AI-agent provenance and hotspots
Bus-factor, per-file maintainer maps, and human/agent collaboration
Architectural decisions extracted from commits and PRs
Ask grounded questions over the indexed code