analyze_change_set

Prompt file imported from codescene-oss/codescene-mcp-server (.github/prompts/analyze_change_set.prompt.md). Copyright stays with the author.

Run a branch-level Code Health analysis using CodeScene MCP Server with analyze_change_set. This is a PR-style safety gate for committed changes on the current branch compared to the given base ref.

After running the tool:

1. Summarize the JSON output in clear, actionable language. Use quality_gates for the pass/fail outcome, but also inspect the top-level metadata object so you do not over-interpret an empty results array.
2. Highlight any code smells or degradation introduced across the branch.
3. Prioritize findings by severity and likelihood of increasing long-term maintenance cost.
4. If regressions are reported, identify which files need follow-up code_health_review work before opening a PR.
5. Provide practical recommendations for improvements before opening a PR.
6. If no regressions are found, state that explicitly and mention any residual risks or testing gaps.
7. When results is empty, explain whether metadata.status indicates checked files with no degradations (no-issues-found) or no applicable branch changes (no-files-modified).

Treat Code Health 10.0 as the ideal long-term target, but use this prompt to prevent regressions and encourage measurable improvement rather than requiring a perfect score on every branch.

Use concise language suitable for a PR pre-flight review. This prompt is a late safety gate, not a replacement for per-file reviews during editing. Avoid rewriting code unless explicitly asked.