incremental-verification — independently scanned and version-tracked by SaferSkills.
SaferSkills independently audited incremental-verification (Agent Skill) and scored it 100/100 (green). The audit ran 55 deterministic rules across Security, Supply Chain, Maintenance, Transparency, and Community; it found 0 high-severity and 0 lower-severity findings. The full rule-by-rule trace and per-finding evidence are below. Free, methodology-open.
Findings & checks · 0 flagged
Every scanned point with the score it earned and what moved between them.
First recorded scan — no prior version to compare against.
The primary manifest — the file an agent reads to learn what this artifact does.
An agent that confirms correctness at every micro-step - after every function, every file change, every integration point - rather than building a large body of code and hoping it works. Without this skill, agents produce code that looks plausible but fails at runtime, and bugs compound silently until the system is too broken to debug efficiently.
After every meaningful change, run a verification:
Write code → Verify → Commit → Next change
↑ |
└────────────────────────────┘Verification methods by change type:
| Change | Verification |
|---|---|
| New function | Run its unit test |
| Modified function | Run existing tests + new test for the change |
| New API endpoint | Send a test request, check response shape and status |
| UI component | Render it, check the DOM/accessibility tree |
| Configuration change | Restart the service, verify it loads without errors |
| Database migration | Run the migration, verify the schema matches expectations |
Before writing new code, confirm the current state is clean:
If any of these fail before you've made changes, flag it immediately. Do not write code on top of a broken foundation.
Write assertions liberally during development:
These can be formal tests or temporary debug assertions that get removed later - the point is catching errors at the moment they occur, not ten steps later.
After every change, verify nothing broke:
CHANGED: src/auth/validate.ts
VERIFICATION STEPS:
1. Run unit tests for auth module → ✅ 12/12 pass
2. Run integration tests touching auth → ✅ 5/5 pass
3. Run full test suite → ✅ 147/147 pass
4. No new linter warnings → ✅
RESULT: Change verified. Committing.If a test fails that you didn't expect, stop. Don't fix it silently. Understand why the change affected that test.
When a task (not just a single change) is complete:
| Failure | Symptom | Correction |
|---|---|---|
| Writing 500 lines then testing | Cascade of errors with unclear root cause | Verify after every function or small logical change |
| Ignoring test failures | Broken tests accumulate; suite becomes unreliable | Fix every failure immediately or investigate why |
| Only testing happy path | Code works for valid input, crashes on edge cases | Test error conditions and boundary values at each step |
| Skipping build verification | Runtime errors that the compiler would have caught | Run build after every file modification |
| Not checking your own diff | Accidental debug code, TODO comments, or unrelated changes ship | Review your diff before marking any task complete |
~30 seconds. Free. No account. Every finding cites a rule and a line of evidence.