kkkirito-123/zlagent/verification-before-completion
Claude Code and agent skill by kkkirito-123/zlagent
- community
Checking that the SKILL.md file is still in the repository.
Do not declare a task done without running verification. For code, that means running tests / smoke; for answers, that means checking the claim against a fresh source. Adapted from obra/superpowers and addyosmani's "Verify" lifecycle phase.
Install
npx vskill@latest install kkkirito-123/zlagent/verification-before-completionSKILL.md
verification-before-completion
Trigger
Activate before saying "done", "fixed", "完成了", "修好了", or shipping any code change to the user. Also activate at the end of every task from task-planning.
Principle
A claim of completion without verification is just optimism. Everydeliverable needs a concrete check that would have caught the bug.
Inputs
task_description: what was being done.artifacts: files edited, commands run, tools called.risk_class:"code" | "data" | "info" | "config"— dictates the
verification depth.
Steps
- Name the check up front (before writing the code) when possible:
"I'll know this is fixed when X."
- For code (`code` class):
- Compile / type-check when available (e.g.
python -m compileall). - Run the relevant test or smoke suite.
- If no suitable test exists, author ONE small targeted test.
- Verify the output / exit code matches the expectation.
- For data transformations (`data` class):
- Spot-check 2-3 representative rows before and after.
- Confirm row counts match expected, or explain the delta.
- For factual claims (`info` class):
- Re-check via an independent fresh
read_urlorweb_search. - Never declare certainty from a single source.
- For config / runtime changes (`config` class):
- Boot / health-check the affected service.
- Exercise the one path the change touches.
- Summarize: what was verified, by what command/check, with what
result.
Verification
- The step that would have caught the bug actually ran.
- The exit code / test result / fresh source is cited explicitly.
- "Done" is never claimed based on "it should work" or "looks right".
Failure Signals
- "Done" with no command output or test pass reference.
- Smoke / compileall skipped because "small change".
- Verification ran against stale cached data.
- The user later reports a regression the verification step would have
caught — audit and tighten the check list.
Output Format
When reporting completion, always include a short verified block:
If any check failed, DO NOT declare done — go back to systematic-debugging.
Security
- Security scanPassed, 90 of 100Pattern checks on the skill's files.
- Model reviewPassed
Quality evaluation
- Task evaluationNot runNo evaluation yet. Run one in Skill Studio.
Source
- SKILL.md in the repositoryCheckingworkspace/skills/coding/verification-before-completion/SKILL.md
Popularity
- Repository stars
- 32
- Repository forks
- 5
- 7-day trend score
- 0
Stars and forks are for the whole GitHub repository, not this skill alone. Not refreshed yet.
Works with
Any agent vskill installs to: 56 targets in all. These 39 have a page on this site.
- Claude Code
- Cursor
- GitHub Copilot
- Windsurf
- Codex
- Gemini CLI
- Cline
- Amp
- Roo Code
- Goose
- Continue
- Kilo Code
- Junie
- Trae
- OpenHands
- Qwen Code
- AI Agent
- Tabnine Chat
- Sourcegraph Cody
- Amazon Q
- Firebase Genkit
- Sweep
- Devin
- Phind
- Mentat
- Claude Desktop
- Aider
- GPT Pilot
- Plandex
- bolt.new
- v0
- Replit Agent
- Anse
- TypingMind
- BetterChatGPT
- LibreChat
- Warp AI
- Pieces
- Fig AI
Embed badge
[](https://verified-skill.com/skills/kkkirito-123/zlagent/verification-before-completion)