From a PR or working-tree diff, list likely regression surfaces (routes, config keys, providers, i18n keys, report templates) and point to existing tests or commands to run. Lightweight pre-review scout — not full QA.
Install
npx skillscat add siinxu/stock-pulse-ai/regression-scout Install via the SkillsCat registry.
Regression Scout
Lightweight surface scout for a change set. Outputs likely regressions and existing tests or recipe commands — it does not replace /test-change or CI.
Source of truth: AGENTS.md §6–7. Recipes: .claude/skills/references/test-command-recipes.md. Reachability: .claude/skills/references/hard-rules.md §3.
Usage
/regression-scout [--base origin/main] [--pr <n>]Instructions
Step 1: Obtain the diff
BASE_REF=$(git merge-base HEAD origin/main)
git diff --name-only "$BASE_REF"..HEAD
# or: gh pr diff <n> --repo SiinXu/stock-pulse-ai --name-onlyStep 2: Classify surfaces
| Diff hits | Scout for |
|---|---|
api/**, src/schemas/** |
API compatibility; OpenAPI drift; Web client types |
src/core/config*, .env.example, config_registry_parts/** |
Settings controls; registry guard; inventory |
data_provider/** |
Priority, timeout, fallback, cache contracts |
templates/**, report services |
Report/notification structure |
apps/dsa-web pages/routes/nav |
Deep links; production reachability |
| i18n / locales | Missing keys; high-risk process |
| workflows / scripts | Gate semantics; path filters |
Step 3: Point to existing tests
rg -n "symbol_or_module" tests apps/dsa-web --glob '*test*' -lMap each surface to a recipe from test-command-recipes.md.
Step 4: Reachability pass
If the diff adds a user-facing panel/API client/service, run hard-rules §3 production-reference grep.
Step 5: Output
Surfaces → commands; reachability verdict; hand off execution to /test-change and merge verdict to /review-pr.
Never
- Claiming the PR is fully verified after scouting only
- Weakening tests
- Auto push / merge