review-impact
Skill4 starsChanged 3 months agoMITView on GitHub
Analyse the blast radius of branch or pull-request changes across a multi-package repo — who breaks if this is wrong, and which test suites must pass. Use before pushing a change to shared code.
1 other project uses a version of this.
Adopted by
Most recently taken up by
- nhhthong/skillstheir own version, first seen Nov 1, 2026
Other versions1 project uses a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 1
- nhhthong/skills★ 1same name
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
In this repository19 other files it gives its agents.
- AGENTS.mdAGENTS.md
- adapter-contract-testingSkill
- azure-devops-prSkill
- bugszero-root-causeSkill
- legacy-3pSkill
- port-refactoringSkill
- quality-gateSkill
- review-ai-writingSkill
- review-allSkill
- review-architectureSkill
- review-complexitySkill
- review-driftSkill
- review-prSkill
- review-securitySkill
- review-slopSkill
- review-testingSkill
- scope-challengeSkill
- teams-local-messagesSkill
- write-testsSkill
The file
Files
Impact Analysis Skill
Maps the blast radius of a change across packages. Two questions: who breaks if this is wrong, and which suites must pass before pushing?
Unlike the other reviewers, this one is most useful before you push, not after. It answers "what did I just put at risk", and its output is a list of commands to run.
Protocol: docs/review-protocol.md — arguments, ref resolution, exhaustiveness, severity levels, report shape. Read it first.
Its § 4 Be exhaustive is mandatory; the lens steps it refers to are steps 2–7 — every changed path bucketed, every modified symbol traced below.
Adapt to your repo. This skill assumes a multi-package repo — shared packages that several deployables depend on. Substitute your own layout for
<shared>/and<app>/, and your own commands for<run tests for X>,<run type checker>,<run linters>. In a single-package repo, steps 4–5 collapse to "grep the whole tree" and the skill is still worth running.
Process
- Resolve refs and gather the diff — per
review-protocol.md. - Bucket every changed path: shared package, deployable, scripts, root config, docs/specs, infrastructure. The interesting ones are shared packages — a change confined to a single deployable has a blast radius of one.
Extract the modified public surface. For each non-test change in a shared package, list what was added, changed or removed: top-level functions, classes, public methods, module constants, settings/config fields, serialized schemas.
For each, classify the change as signature or behavioural — a changed return type, a new error path, a different side effect, a changed default. This is the single most important distinction in the report: signature breaks are caught by the compiler or type checker, behavioural breaks are caught in production.
Trace callers. For each symbol above, search the whole repo:
grep -rn '\b<symbol>\b' --include='<source-glob>' .Group hits by package and by kind — same-package, sibling-shared-package, deployable-runtime, deployable-test. Track re-exports separately: a symbol re-exported from a package entry point has a wider surface than its definition suggests.
- Trace dependency changes. For each modified manifest, list added, removed and bumped dependencies, then find which deployables depend on that package.
- Trace the cross-cutting cases. These are the ones that bite, because the compiler
sees nothing:
- Config/settings classes — renaming or removing a field is an environment-contract change. List every consumer, and remember the deployment configs outside this repo.
- New test infrastructure (shared fixtures, conftest, test base classes) — often needs a full type-check rather than an incremental one.
- Message/queue contracts and event schemas — list every producer and consumer. A rolling deploy means old and new run simultaneously; ask whether the change is compatible in both directions.
- Database models and migrations — list every service reading or writing the same table or collection.
- Recommend verification. Concrete, copy-pasteable commands, one per impacted package. The point of the report is this list.
- Confirm § 4 of the protocol is satisfied, then generate report →
reviews/impact-analysis.md
Severity Model
| Severity | Indicators |
| Critical | Behavioural change to a symbol with at least one runtime caller outside its package; config field renamed or removed; message/schema contract change |
| Major | Signature change with multiple callers; dependency bump affecting two or more deployables; re-exported symbol modified |
| Minor | Purely additive surface; change confined to one package; test-only callers |
Output Format
# Impact Analysis: {branch}
## Change Summary
- Shared packages touched: …
- Deployables touched: …
- Public symbols modified: N (M behavioural, K signature-only)
## Blast Radius
### <shared>/pkg — `Foo.do_thing()` (behavioural)
**Callers (5):**
- <app-a>/…/bar.py:42 (runtime)
- <app-b>/…/baz.py:13 (runtime)
- <shared>/other/qux.py:88 (sibling)
**Risk:** Critical — the error path changed and the runtime callers don't handle the new exception.
**Required:** `<run tests for app-a>`, `<run tests for app-b>`, `<run tests for shared/other>`
## Cross-Cutting Concerns
- `Settings.broker_url` renamed → environment contract change; update deployment configs
- New shared fixture → run the full type check, not the incremental one
## Recommended Verification
```bash
<run tests for app-a> <run tests for app-b>
<run type checker>
<run linters>Summary
- Critical: 1 | Major: 2 | Minor: 0
- Push-blocking: yes/no
## Example Usage
review-impact # Current branch vs main review-impact feat/foo # Specific branch review-impact feat/foo dev # Different base review-impact 3574 # PR/MR by number
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

