Codebase intelligence for AI and humans: code health scores, auto-generated docs, git analytics, dead code detection, and architectural decisions via MCP.
Python aicode-complexitycode-healthcode-intelligencecode-qualitydead-codedeveloper-toolsdocumentationgit-analyticsmcpopen-sourcepythonrefactoringstatic-analysistechnical-debt
Claude Code Slash Commands
The Repowise Claude Code plugin ships 18
slash commands under the /repowise: namespace. Install the plugin with
/plugin install repowise@repowise (see the plugin README), then run
/repowise:init to index the repo.
How they work
- Each command is a Markdown instruction file in
plugins/claude-code/commands/. Claude reads it, runs the matchingrepowiseCLI command, and presents the result. - The file's
allowed-tools:frontmatter scopes what Claude may run: each command allows only its own CLI subcommand (for exampleBash(repowise health:*)) plusRead. /repowise:initalso allows the install commands (uv tool install,pipx install,pip install) andAskUserQuestion, so it can install repowise and ask about your setup./repowise:reindexalso allowsAskUserQuestion.- Text after the command name reaches the command as
$ARGUMENTS:
/repowise:search authentication middleware
/repowise:context src/auth/token.py
/repowise:dead-code safe
/repowise:risk main..HEAD
Command reference
| Command | What it does |
|---|---|
/repowise:init |
Install repowise if needed, ask about your preferences, and index the repo |
/repowise:status |
Index health: sync state, page counts, provider, token usage |
/repowise:update |
Incremental update to bring the index up to recent commits |
/repowise:search |
Search the wiki by natural language, full text, or symbol |
/repowise:ask |
Cited answer to a codebase question with a confidence rating (one LLM call) |
/repowise:context |
Triage card for files, modules, or symbols: layer, hotspot, fix history, freshness |
/repowise:symbol |
One symbol's body with live-verified line bounds (path::Name, a line range, or a distill omission ref) |
/repowise:reindex |
Re-embed every wiki page into the vector store. Embedding calls only, no LLM calls |
/repowise:health |
Code health: KPIs, lowest-scoring files, refactoring targets, trends, per-file markers |
/repowise:coverage |
Ingest or inspect coverage reports (LCOV, Cobertura, Clover, Go, JaCoCo, Apex, repowise JSON; a coverage.py .coverage file builds the per-test map), or gate a change on patch coverage |
/repowise:impacted-tests |
Tests whose coverage touches a change's lines, for a commit, a base..head range, or the staged diff |
/repowise:risk |
Rank a change against recent history: repo-relative percentile plus a supporting diff-shape score |
/repowise:security |
Security signals. Working-tree scanning runs during init and update; --history walks the full git history for leaked secrets |
/repowise:dead-code |
Unreachable files, unused exports, and zombie packages, tiered by confidence |
/repowise:export |
Export the wiki or architecture model to Markdown, HTML, JSON, or Structurizr DSL |
/repowise:decision |
List, inspect, add, or confirm architectural decisions |
/repowise:why |
Why the code is shaped this way: decisions, rationale, git archaeology |
/repowise:doctor |
Diagnose the setup (install, API keys, index and store drift) and optionally repair it |
Slash commands and MCP tools
- Slash commands are for you: run one when you want a specific report, such as dead code before a cleanup or risk before a merge.
- MCP tools are for Claude: it calls them on its own while working, for
example
get_riskbefore editing a hotspot. See MCP_TOOLS.md.
Both read the same .repowise/ index. The command files add presentation
rules on top of the CLI output, such as showing the confidence with every
/repowise:ask answer and treating only safe_to_delete findings as
deletions in /repowise:dead-code.