Skip to main content

gitnexus-impact-analysis

Check dependency and execution-flow impact for shared behavior, interface changes, or an explicit impact-analysis request.

Jump to install

Source facts

Repository
Gracker/SmartPerfetto
Last source activity
September 14, 2026 at 02:16
Detected SKILL.md language
English
Stars
700
Forks
119

Install options

The review-first prompt is selected by default. You can switch to a direct command or download a local copy.

Review the source files

Read SKILL.md and any companion files shown by SkillsMP before deciding whether to install.

Showing SKILL.md

SKILL.md
Source instructions · Read-only preview
name
gitnexus-impact-analysis
description
Check dependency and execution-flow impact for shared behavior, interface changes, or an explicit impact-analysis request.
# Impact analysis with GitNexus Follow the current project's scope and verification rules. Use graph analysis for shared behavior, dependency, interface, or module-boundary changes, or an uncertain cross-module bug. Docs, comments, and small local edits preserving behavior and dependencies do not need graph analysis merely because a symbol is touched. Use `impact` on the key changed symbol to find upstream dependencies. Review direct callers first and read relevant source to determine compatibility: ```text impact({target: "symbolName", direction: "upstream", minConfidence: 0.8, maxDepth: 3}) ``` A direct dependency is not proof of breakage. Judge risk by the actual behavior and contracts changed, not a fixed symbol count. Report HIGH or CRITICAL graph results and cross-check them before the dependent edit; do not dismiss a risk that remains unexplained. Read a specific affected process or use `context` only when its call chain resolves a remaining question. Do not load every process by default. Before committing changes in this scope, inspect task-owned staged changes: ```text detect_changes({scope: "staged"}) ``` Preserve unrelated dirty changes. Reuse relevant graph results when the source and index have not changed. Refresh an index only when its missing or stale coverage matters, using the project's runner with `analyze --index-only` so maintained agent instructions are preserved. Do not run setup as a prerequisite. If the graph is unavailable or remains unreliable after a justified refresh, use direct references, source and affected tests; report uncertainty and resolve material impact before proceeding. Stop once the task's impact is understood.
View on GitHub