From harness-claude
Graph-based impact analysis — answers "if I change X, what breaks?"
npx claudepluginhub intense-visions/harness-engineering --plugin harness-claude[--path <path>] [--files <files>]commands/<!-- Generated by harness generate-slash-commands. Do not edit. --> <context> Cognitive mode: analytical-reporter Type: rigid </context> <objective> Graph-based impact analysis — answers "if I change X, what breaks?" Phases: - identify: Identify changed files from diff or input - analyze: Query graph for impact of each changed file - assess: Rank findings by risk and generate report </objective> <execution_context> @agents/skills/claude-code/harness-impact-analysis/SKILL.md @agents/skills/claude-code/harness-impact-analysis/skill.yaml </execution_context> <process> 1. Try: invoke mc...
/what-would-breakAnalyzes blast radius of proposed code changes via dependency mapping, risk assessment, and mitigation strategies to teach impact analysis.
/impact-analysisAnalyzes impact of changes to a target (file/function/class/description) via interactive context questions, producing layered Markdown reports on affected files, risks, and recommendations in .dcs/ directory.
/SKILLAnalyzes impact of changes to a target (API/model/component) via static dependency analysis, listing affected code, indirect dependencies, tests, and migration checklist.
/vcsdd-coherence-impactRuns BFS forward-impact analysis from changed nodes through the CEG, surfacing CoDD must_review alerts alongside Green/Amber/Gray downstream impact bands.
/impactMaps blast radius before changing a file or function: importers, tests, CI/CD pipelines, docs references, owners via 5 parallel agents.
/impactPerforms cascade impact analysis on a file by reading module docstrings to show dependency chain of callers up to 3 levels, with exports, rules, and summary.
Share bugs, ideas, or general feedback.
Phases:
<execution_context> @agents/skills/claude-code/harness-impact-analysis/SKILL.md @agents/skills/claude-code/harness-impact-analysis/skill.yaml </execution_context>
1. Try: invoke mcp__harness__run_skill with skill: "harness-impact-analysis" 2. If MCP unavailable: read SKILL.md and follow its workflow directly 3. Pass through any arguments provided by the user