agentleFS
Sign inStart free
ovargasovargas/virtual-team/skills/verification-before-completion/SKILL.md

verification-before-completion

Skill14 starsChanged 8 months agoMITView on GitHub

Use before making any completion claim — enforces evidence-before-claims with no exceptions

189 other projects use a version of this, 1 of them edited.

Adopted by

Most recently taken up by

Used with

Projects with it also use chinese-code-review, chinese-commit-conventions and 10 more.

Other files people use alongside this one.

Files found in the same public projects as a version of this one, more often than they turn up anywhere else. Each project counts once, large collections count for less, and this file's own repository is left out.
  • Leodorareluctant259Skill

    chinese-code-review

    中文代码审查规范——在保持专业严谨的同时，用符合国内团队文化的方式给出有效反馈

    Also in 8 projects that use this

  • Leodorareluctant259Skill

    chinese-commit-conventions

    中文 Git 提交规范 — 适配国内团队的 commit message 规范和 changelog 自动化

    Also in 8 projects that use this

  • obraSkill

    diagnosing-superpowers

    Use when a superpowers session went wrong and your human partner wants to know why — repeated work, ignored plans, stumbles, poor results, a skill that didn't fire, "it took too long", "why is it so expensive", "what is it doing" — or wants to build a bug report for the superpowers maintainers, for the current session or a past one identified by id or path, on any harness.

    Also in 20 projects that use this

  • jh941213Skill

    docs-ci

    문서 CI(링크·OpenAPI lint·breaking 게이트·mermaid 검증)나 문서 드리프트 감지(docs.yaml 신선도 검사)를 설치하거나 돌릴 때 사용. 문서 내용 작성은 docs-architecture·docs-interfaces·docs-manuals.

    Also in 3 projects that use this

  • jh941213Skill

    docs-interfaces

    API·이벤트 인터페이스 문서(OpenAPI·swagger·AsyncAPI 명세, 엔드포인트 레퍼런스, API 구성도·시퀀스, API CHANGELOG)를 만들거나 코드와 맞출 때 사용. 시스템 구성도는 docs-architecture, API 설계 원칙은 api-design-principles.

    Also in 3 projects that use this

  • +3 more from Leodorareluctant259
  • +4 more from jh941213

The file

We split this skill into the rules it gives an agent

8 rules. 0 are common, 8 are distinctive, and 0 are very specific to this project.

  • 0

    Common

    Most of the 177 edited versions we read keep these.

  • 8

    Distinctive

    This author's own choices. Worth a look.

  • 0

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    No completion claims without fresh verification evidence.

    line 10
  2. Distinctive:

    This skill is rigid: Follow it exactly. Do not adapt, skip, or soften the discipline.

    line 12
  3. Distinctive:

    Before the agent says ANYTHING positive about work status — "done", "tests pass", "looks good", "fixed", "complete" — it must pass the gate …

    line 22
  4. Distinctive:

    Every positive claim about work status must pass through these 5 steps, in order:

    line 26
  5. Distinctive:

    Run it NOW. Not "I ran it earlier." Not "it passed last time." Fresh means in this message exchange.

    line 34
  6. Distinctive:

    Read the ENTIRE output. Check the exit code.

    line 38
  7. Distinctive:

    If any of these are true, you are about to make an unverified claim:

    line 82
  8. Distinctive:

    Exception: If a verification command was run 2-3 tool calls ago in the SAME message and NO code was modified since, the evidence is still …

    line 99

“Common” is measured: kept by at least 6 in 10 of the 177 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Verification Before Completion

Overview

No completion claims without fresh verification evidence.

This skill is rigid. Follow it exactly. Do not adapt, skip, or soften the discipline.

"Claiming work is complete without verification is dishonesty, not efficiency."

The Iron Rule

NO COMPLETION CLAIMS WITHOUT FRESH VERIFICATION EVIDENCE

Before the agent says ANYTHING positive about work status — "done", "tests pass", "looks good", "fixed", "complete" — it must pass the gate function below.

The Gate Function

Every positive claim about work status must pass through these 5 steps, in order:

1. IDENTIFY — What command proves this claim?

Name the exact command. "Tests pass" → npm test or pytest. "Build succeeds" → npm run build. No vague references.

2. RUN — Execute the command fresh

Run it NOW. Not "I ran it earlier." Not "it passed last time." Fresh means in this message exchange.

3. READ — Full output, check exit code, count failures

Read the ENTIRE output. Check the exit code. Count failures, errors, warnings. Don't skim. Don't assume.

4. VERIFY — Does output actually confirm the claim?

Does the output ACTUALLY say what you're about to claim? "0 failures" confirms "tests pass." "exit 0" confirms "build succeeds." Partial output does not confirm total success.

5. ONLY THEN — Make the claim, citing the evidence

Now you may claim. Cite the evidence: "Tests pass (24 passed, 0 failed)." "Build succeeded (exit 0, no warnings)."

If any step fails or is skipped, the claim cannot be made.

Claim-Evidence Table

ClaimRequired EvidenceNot Sufficient
"Tests pass"Test command output showing 0 failures in this messagePrevious run, "should pass", partial suite
"Build succeeds"Build command with exit code 0Linter passing, "compiled without errors"
"Bug fixed"Reproduction test now passes (run fresh)"Code changed, should work now"
"Phase complete"All phase verification checks green (shown above)"Code looks correct to me"
"PR ready"Tests + lint + type check all run fresh"I verified earlier in the session"
"Linting clean"Lint command output showing 0 errors"No lint rules violated" without running
"Type check passes"Type checker output with 0 errors"Types look correct"
"Feature works"Demo or test proving the feature operates as specified"I followed the spec correctly"

Rationalization Prevention

These thoughts mean STOP — you're about to skip verification:

ThoughtReality
"I just ran it"Was it in THIS message? If not, it's stale. Run again.
"The code is obviously correct"Obvious code fails. Run the proof.
"Nothing changed since last run"Something always changes. Re-run costs seconds.
"Tests would catch it"Did you run them? Show the output.
"I'm confident this works"Confidence is not evidence. Run the command.
"It's just a small change"Small changes cause big failures. Verify.
"Verification will slow me down"False completion claims waste MORE time. Verify.
"The founder can test it"Your job is to deliver verified work, not unverified drafts.
"I verified a similar thing earlier"Similar is not same. This claim needs this evidence.
"Let me just commit and we'll see"No. Verify before commit, not after.

Red Flags — STOP Before Claiming

If any of these are true, you are about to make an unverified claim:

  • Using "should", "probably", "seems to" about work status
  • Saying "Great!", "Perfect!", "Done!" before running verification
  • About to commit, push, or create a PR without fresh test output
  • Claiming phase-complete based on visual code review alone
  • Expressing satisfaction before checking exit codes
  • Referencing a verification run from earlier in the conversation as current proof
  • Saying "all tests pass" without test output visible in this message
  • Claiming "no errors" without having run the checker

Any of these mean: STOP. Run verification. Show output. Then claim.

What "Fresh" Means

  • Fresh evidence = command was run AND output was read in the current message exchange
  • Stale evidence = output from a previous message, a previous phase, or a previous session
  • Exception: If a verification command was run 2-3 tool calls ago in the SAME message and NO code was modified since, the evidence is still fresh. The rule targets recall across message boundaries, not ritual re-runs within the same step.

Integration

This skill is loaded by:

  • /virtual-team:implement — Layer 0 (behavioral discipline), loaded alongside virtual-team:test-driven-development
  • /virtual-team:commit — Before committing, verify tests pass fresh
  • /virtual-team:pr — Before creating PR, full verification suite run fresh
  • /virtual-team:debug — Before claiming "root cause found", verify hypothesis with evidence
  • /virtual-team:review — Before issuing verdict, verify claims about code behavior
  • /virtual-team:flow — inherited through /virtual-team:implement

Download now (ZIP)

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
What did you use it for? e.g. Reviewing pull requests in a Rails app

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.

Check your setup, free