From mz-dev-pipe
ALWAYS invoke when the user wants to see the impact of changing a file, function, or module before refactoring. Triggers: "blast radius of X", "what depends on X", "impact analysis". When NOT to use: the refactor itself (use optimize or build).
npx claudepluginhub doctormozg/claude-pipelines --plugin mz-dev-pipeThis skill is limited to using the following tools:
Orchestrates a read-only impact analysis pipeline. Given a target (file, function, or module), computes the full dependency graph, overlays git age data, scores risk, and produces a ranked report showing what breaks if the target changes.
Searches, retrieves, and installs Agent Skills from prompts.chat registry using MCP tools like search_skills and get_skill. Activates for finding skills, browsing catalogs, or extending Claude.
Checks Next.js compilation errors using a running Turbopack dev server after code edits. Fixes actionable issues before reporting complete. Replaces `next build`.
Guides code writing, review, and refactoring with Karpathy-inspired rules to avoid overcomplication, ensure simplicity, surgical changes, and verifiable success criteria.
Share bugs, ideas, or general feedback.
Orchestrates a read-only impact analysis pipeline. Given a target (file, function, or module), computes the full dependency graph, overlays git age data, scores risk, and produces a ranked report showing what breaks if the target changes.
optimize or build.debug.audit.$ARGUMENTS — A file path, function name, or module name. If empty or ambiguous, ask via AskUserQuestion. Never guess.See skills/shared/scope-parameter.md for the canonical scope modes (branch, global, working) and their git commands. Document any skill-specific overrides or restrictions below this line.
$ARGUMENTS); scope narrows only the graph-expansion file set, not the grep-for-references universe. Cross-repo references must still be surfaced if they touch the target.scope:): all project files are eligible for reference discovery..mz/task/| # | Phase | Reference | Loop? |
|---|---|---|---|
| 0 | Setup | inline below | — |
| 1 | Discovery & Graph | phases/discovery.md | hops |
| 2 | Analysis & Report | phases/analysis_and_report.md | — |
$ARGUMENTS to determine target type:
/ or . extension → validate file exists via Glob.<YYYY_MM_DD>_blast_radius_<slug> where <YYYY_MM_DD> is today's date (underscores) and slug is a snake_case summary (max 20 chars) of the target basename; on same-day collision append _v2, _v3. Create .mz/task/<task_name>/. Write state.md with Status: started, Phase: setup, Target, Target type, Started timestamp. Use TaskCreate for tracking.Dispatch MAX_RESEARCHERS (4) pipeline-researcher agents (model: sonnet) in parallel, each covering a distinct reference category. Then run iterative hop expansion up to MAX_DEPTH.
See phases/discovery.md for researcher dispatch prompts, reference categories, hop expansion algorithm, and graph.md artifact.
Update state phase to graph_complete.
Overlay git age data on every node in the graph. Compute risk scores. Generate the final report with safety verdict.
See phases/analysis_and_report.md for age analysis commands, risk scoring formula, report template, and safety verdict criteria.
Write report to .mz/reports/<YYYY_MM_DD>_blast_radius_<target_slug>.md. Present summary to user. Update state to completed.
Techniques: delegated to phase files — see Phase Overview table above.
| Rationalization | Rebuttal |
|---|---|
| "small refactor, skip the map" | "small refactors produce the most silent breakage because reviewers don't look" |
| "I know what this touches" | "you know what the call graph you remember touches, not what it actually touches" |
| "the tests will catch it" | "tests only catch what they cover; the map catches what tests miss" |
Output the final report block: target, graph node count, top risk dependents, safety verdict, and the written report path.
grep -r "<target basename>" --include="*.{js,ts,py,go,rs,java}" . or equivalent for the target type). If the smoke test also returns no matches, emit ZERO RESULTS VERIFIED in the report and proceed to Phase 2 (Analysis & Report). If the smoke test returns matches (indicating a search gap in the researcher dispatch), escalate via AskUserQuestion before continuing — do not proceed to reporting.Update .mz/task/<task_name>/state.md after each phase with: current phase, graph node count, files analyzed, risk summary. All intermediate artifacts persist in the task directory.
Produce a ranked report showing exactly what depends on the target and how risky each dependency is. Every claim must trace back to a specific grep match or git log entry — never fabricate references.