Agent skill
gate
Verification before completion claims. Use when about to say "done", "fixed", or "complete". Runs checklist by workflow type with evidence requirements.
Install this agent skill to your Project
npx add-skill https://github.com/majiayu000/claude-skill-registry/tree/main/skills/data/gate
SKILL.md
gate
Verify before claiming completion.
Trigger
Before claiming: "done", "fixed", "complete", "working", "ready"
Pre-Work Gate (Before Starting)
Run before substantial work begins. Blocks execution if missing.
□ Verification method locked (not "assumption")
→ What type: execution output / observation / measurement?
→ What evidence will you show?
□ At least 2 "must NOT" criteria exist
→ What should this NOT do?
→ What scope should NOT expand?
□ Fit score determined workflow shape
→ <25: BLOCKED - clarify first
→ 25-29: Colleague-shaped (iterate each step)
→ 30-39: Tool-with-review (checkpoint major steps)
→ 40+: Tool-shaped (milestones only)
If any missing: STOP. Complete before proceeding.
Post-Work Gate (Before Claiming Done)
Workflow A: Build
□ Feature executes without errors (show output)
□ Edge cases tested (empty, null, boundary)
□ Rollback tested if Type 2B/1
□ Dependencies verified
Workflow B: Debug
□ Root cause identified with evidence
□ Fix tested and resolves symptom
□ Prevention added (test/automation/guard)
□ Related bugs checked (same class)
Workflow C: Refactor
□ Existing tests pass unchanged
□ Behavior identical (output comparison)
□ Deletion complete (no orphans)
□ No new abstraction without justification
Verification Types
| Type | Description | Sufficient for SHIP? |
|---|---|---|
execution output |
Ran command, showed result | ✓ Yes |
observation |
Screenshot, debugger session | ✓ Yes |
measurement |
Metrics, benchmark data | ✓ Yes |
reasoned inference |
Logic-based conclusion | ⚠️ Flag, don't block |
code review |
Inspection only | ⚠️ Weak |
assumption |
Not verified | ⚠️ Flag with warning |
Default: Require execution output, observation, or measurement before completion claims.
Boundary
Gates advise, never prevent. User owns their work.
- If user says "I'm confident" → flag concern, proceed
- If user says "skip gates" → proceed without verification demands
- Never invalidate user's assessment of their own work
Gate exists to surface risk, not to block legitimate intuition.
Verification Plan
Before claiming completion, define how you'll verify:
| Verification Type | Evidence Required | Time Cost |
|---|---|---|
| Execution output | Command output, test results | < 1 min |
| Observation | Screenshot, debugger inspection | 1-5 min |
| Measurement | Metrics, benchmarks, profiling | 5-15 min |
| Code review | Manual inspection only | Variable |
| Assumption | No verification | 0 (blocks SHIP) |
Verification Requirements by Stakes
| Stakes | Minimum Verification |
|---|---|
| High | Execution output + Observation |
| Medium | Execution output |
| Low | Code review acceptable |
Anti-Footgun Rules
- Never claim "done" with only assumptions
- "It compiles" is not verification
- "Tests pass" requires showing test output
- If verification would take > 15 min, that's a smell (task too big)
Anti-Patterns
| Claim | Problem | Required |
|---|---|---|
| "Should work" | Speculation | Actual output |
| "Looks good" | No verification | Test results |
| "Fixed the issue" | No proof | Before/after |
| "I think this resolves it" | No confidence % | X-Y% + basis |
Incomplete Work
When blocked, state:
- What's done (with evidence)
- What's blocking (specific)
- What's remaining
Type 2A Exception
Trivial changes (< 1 min rollback): completion allowed without full gate. Document undo command.
Common Rationalizations (All Wrong)
| Thought | Reality |
|---|---|
| "The tests passed" | Tests ≠ verification. Did you run the ACTUAL command? |
| "It should work" | "Should" = assumption. Show evidence. |
| "I already checked" | When? Show the output. |
| "The user is waiting" | Rushed verification = rework later. |
| "It's a simple change" | Simple changes break production too. |
| "I'm pretty sure it works" | Pretty sure ≠ evidence. Run it. |
Recommended Agent Skills
Expand your agent's capabilities with these related and highly-rated skills.
majiayu000/claude-skill-registry
agent-ops-spec
Manage specification documents in .agent/specs/. Use when user provides requirements, acceptance criteria, or feature descriptions that need to be tracked and validated against implementation.
majiayu000/claude-skill-registry
agent-ops-state
Maintain .agent state files. Use at session start, after meaningful steps, and before concluding: read/update constitution/memory/focus/issues/baseline consistently.
majiayu000/claude-skill-registry
agent-ops-spec
Manage specification documents in .agent/specs/. Use when user provides requirements, acceptance criteria, or feature descriptions that need to be tracked and validated against implementation.
majiayu000/claude-skill-registry
agent-ops-testing
Test strategy, execution, and coverage analysis. Use when designing tests, running test suites, or analyzing test results beyond baseline checks.
majiayu000/claude-skill-registry
agent-ops-testing
Test strategy, execution, and coverage analysis. Use when designing tests, running test suites, or analyzing test results beyond baseline checks.
majiayu000/claude-skill-registry
agent-ops-state
Maintain .agent state files. Use at session start, after meaningful steps, and before concluding: read/update constitution/memory/focus/issues/baseline consistently.
Didn't find tool you were looking for?