Przejdź do treści
Biblioteka skilli
DevOps i chmura · Testowanie

verification-before-completion

/verification-before-completion

Wymaga sprawdzenia infrastruktury przed uznaniem jej za gotową, zdrową lub zweryfikowaną, przed commitem i aktualizacją statusu.

Repozytorium
srepowers ↗
Licencja
MIT
Ostatnia zmiana repozytorium
09.11.2026

Ocena i bezpieczeństwo

100/ 100Klasa: dobry
Format25 / 25
Frontmatter z nazwą i opisem, poprawna nazwa, opis do 1024 znaków, nazwa zgodna z folderem.
Odkrywalność25 / 25
Opis mówi, kiedy użyć skilla, ma sensowną długość i jest w 3. osobie.
Treść35 / 35
Co najmniej 150 słów i do 500 linii, nagłówki, listy kroków, przykłady albo kod.
Higiena15 / 15
Bez zaślepek typu TODO, bez ścieżek Windows, z licencją.
Ryzykobrak sygnałów w skanie
Pliki w paczce
2
Skrypty
0
Komentarze HTML
0
Adresy URL w SKILL.md
1
Przestarzałość

Brak sygnałów. Repozytorium zmienione 09.11.2026, bez odwołań do wycofanych modeli i API.

Licencja

MIT, otwarta: można używać i zmieniać, z zachowaniem informacji o autorze.

Co robi i kiedy

Wymaga sprawdzenia infrastruktury przed uznaniem jej za gotową, zdrową lub zweryfikowaną, przed commitem i aktualizacją statusu.

Oryginalny opis (EN)

Use when about to claim infrastructure work is complete, fixed, verified, or healthy - before any commit, PR, or status update

Jak działa

  1. Overview
  2. The Iron Law
  3. Why This Matters
  4. The Gate Function
  5. Project-Wide Validation
  6. Common Infrastructure Failures
  7. Red Flags — STOP
  8. Rationalization Prevention
  9. Key Patterns
  10. When To Apply
Sekcje z nagłówków SKILL.md, w oryginale · 179 linii, 1 157 słów
Podgląd SKILL.md
---
name: verification-before-completion
description: Use when about to claim infrastructure work is complete, fixed, verified, or healthy - before any commit, PR, or status update
license: MIT
---

# Verification Before Completion

## Overview

Claiming infrastructure work is complete without verification causes incidents, not efficiency.

**Core principle:** Evidence before claims, always.

**Violating the letter of this rule is violating the spirit of this rule.**

## The Iron Law

```
NO COMPLETION CLAIMS WITHOUT FRESH VERIFICATION EVIDENCE
```

If you haven't run the verification command in this message, you cannot claim it passes.

## Why This Matters

Unverified completion claims are how single changes become incidents:
- "It's deployed" when the rollout never became Ready — traffic hits a broken version
- Undefined config shipped — the service crash-loops on next restart
- Missing requirements shipped — the change is incomplete and reopens as a follow-up incident
- Time wasted on false completion → the operation gets handed back → rework under worse conditions
- Trust broken — once a stakeholder hears "it's fixed" and finds it isn't, every future status claim is doubted
- Violates the core value: honesty. A false "healthy" during an incident is worse than saying "still investigating."

## The Gate Function

```
BEFORE claiming any status or expressing satisfaction:

1. IDENTIFY: What command proves this claim?
2. RUN: Execute the FULL command (fresh, complete)
3. READ: Full output, check exit code, count failures
4. VERIFY: Does output confirm the claim?
   - If NO: State actual status with evidence
   - If YES: State claim WITH evidence
5. ONLY THEN: Make the claim

Skip any step = lying, not verifying
```

## Project-Wide Validation

When work changes repository content, the project's complete validation command
or full suite defines green — not only the one test, manifest, or module named
in the task. A scope statement bounds the deliverable, not verification.

- Run the repository's canonical complete command (`pytest`, `npm test`,
  `cargo test`, `python3 scripts/validate-repo.py`, or its documented equivalent)
- Report every failure by name, including failures you did not cause; omission
  falsifies the completion report
- Distinguish pre-existing failures from regressions with evidence, but do not
  silently suppress either and do not expand scope to fix unrelated failures
- A targeted check may establish RED/GREEN for the change; it does not replace
  the complete project gate before a completion, commit, PR, or merge claim

For live-state-only operations with no repository suite, the complete command is
the plan's full verification set for the affected and adjacent systems.

## Common Infrastructure Failures

| Claim | Requires | Not Sufficient |
|-------|----------|----------------|
| Deployment succeeded | `kubectl --context <context> rollout status` shows all ready | Helm exit 0, "should be up" |
| Service is healthy | Health endpoint returns 200 + correct response | Pod is Running |
| Config applied | `kubectl --context <context> get cm/secret -o yaml` shows new value | `kubectl --context <context> apply` exit 0 |
| Certificate valid | `openssl s_client` or cert expiry checked | Secret created |
| Migration complete | Row count / schema verified | Migration job exit 0 |
| Pipeline passed | CI job logs show green + artifacts | Job queued or "triggered" |
| Bug fixed | Original symptom reproduced and no longer fails | Code changed |
| Rollback complete | Service responding correctly at previous version | Old image tag applied |
| Resource deleted | `kubectl --context <context> get` returns NotFound | Delete command exit 0 |
| Agent completed task | VCS diff shows expected changes | Agent reports "success" |

## Red Flags — STOP

- Using "should", "probably", "seems to", "looks like"
- Expressing satisfaction before verification ("Done!", "Fixed!", "Deployed!", "All good!")
- About to commit/push/PR/merge without verification
- Trusting agent success reports without independent check
- Relying on command exit code alone (exit 0 ≠ correct result)
- Thinking "just this once" during an incident
- Tired at end of incident and wanting work over
- **ANY wording implying success without having run verification**

## Rationalization Prevention

| Excuse | Reality |
|--------|---------|
| "Helm exited 0, it's deployed" | Run `kubectl --context <context> rollout status` |
| "I'm confident it's fixed" | Confidence ≠ evidence. Run the check. |
| "Pod is Running" | Running ≠ healthy. Check health endpoint. |
| "Agent said success" | Verify independently with kubectl/git diff |
| "I'm tired, incident is over" | Exhaustion ≠ excuse. One more command. |
| "Partial check is enough" | Partial proves nothing. Run the full check. |
| "It worked in staging" | Staging ≠ production. Verify production. |
| "No time during incident" | Unverified fix extends incidents. |

## Key Patterns

**Deployment:**
```
✅ [kubectl --context <context> rollout status deploy/api -n prod] [See: successfully rolled out] "Deployment succeeded"
❌ "Helm exited 0, deployment done"
```

**Service health:**
```
✅ [curl -s https://api.example.com/healthz] [See: {"status":"ok"}] "Service healthy"
❌ "Pod is Running"
```

**Config/Secret applied:**
```
✅ [kubectl --context <context> get cm app-config -o jsonpath='{.data.KEY}'] [See: expected-value] "Config applied"
❌ "kubectl --context <context> apply returned exit 0"
```

**Infrastructure operation (TDO):**
```
✅ [Run verification command from plan] [See: expected output] "Phase complete"
❌ "Commands ran without errors"
```

**Agent delegation:**
```
✅ Agent reports success → [git diff / kubectl --context <context> get] → Verify changes → Report actual state
❌ Trust agent report at face value
```

## When To Apply

**ALWAYS before:**
- ANY success/completion/healthy claim
- ANY expression of satisfaction
- Committing, PR creation, task completion
- Moving to next step in an operation plan
- Handing off to next skill (e.g., finishing-operation-branch)
- Delegating to subagents and accepting their results
- Closing an incident

**Rule applies to:**
- Exact phrases and paraphrases
- Implications of success
- ANY communication suggesting completion or correctness

## Integration

**Pairs with:**
- **test-driven-operation** — TDO's verification step IS this skill's gate function
- **executing-operation-plans** — Each phase completion requires verification before proceeding
- **finishing-operation-branch** — Branch completion requires all verifications passed
- **incident-commander** — Incident close requires verified service restoration

## Requirements Traceability Verification

When verifying completion of an operation plan that includes a `## Requirements Traceability` section:

1. Read the requirements traceability table from the plan
2. For each requirement mapped to a task, verify the task's execution evidence covers it
3. Update the Status column: `pending` → `done` | `skipped` | `failed`
4. Present a coverage summary:

```
Requirements Coverage:
✅ "Pod has resource limits" → Task 3 → done
✅ "Health check configured" → Task 3 → done
❌ "Network policy applied" → Task 4 → failed (verification output mismatch)
```

5. If any requirement shows `failed` or is still `pending`: the operation is NOT complete
6. All requirements must be `done` or explicitly `skipped` (with documented reason) before proceeding to `finishing-operation-branch`

**Issue tracker integration:** When the plan has a `ticket` in frontmatter, cross-reference acceptance criteria against the coverage summary. If a tracker MCP (ClickUp, Jira, etc.) is available, fetch the ticket and verify criteria are met. Update the ticket to reflect completion when all criteria pass.
Pliki w paczce (2)
  • README.md 2 KB
  • SKILL.md 8 KB

Wersje

RepozytoriumWynikLicencja
Ta strona · yg-codes/srepowers100MIT
jjuraszek/pi-gauntlet100MIT
GanyuanRan/Aegis100MIT
mvschwarz/openrig100Apache-2.0
nikospavlopoulos/devops-mutant-powers100MIT
FrancoStino/opencode-skills-collection100MIT
foryourhealth111-pixel/Vibe-Skills100Apache-2.0
rootcastleco/rei-skills100MIT
Montinou/triqual100MIT

Migawka bazy z 13.11.2026 · wersja paczki c7f2a56b42dd (skrót SHA-256 całej paczki)

Pobrania w ostatnich 30 dniach: 0 · ranking

Zgłoś problem z tym skillem: szkodliwa treść, błędny opis albo prośba autora o usunięcie.

Podobne skille

Uwagi i zgłoszenia

Wiadomość trafi do Bartosza Sokolińskiego. Adres e-mail jest potrzebny tylko do odpowiedzi. Szczegóły w polityce prywatności.