- Home
- Skills
- Testing & Debugging
- Implementation Verification
Works with the AI Agents you already use
Implementation Verification
Proves a change actually works by running the real checks — never claims "verified" based on what a passing local run merely suggests.
$4.99
Implementation Verification
Example session with this skill installed
I just added visual regression tests and they pass locally in Docker — can you verify before I call this done?
- Read your context and instructions
- Compiled the implementation verification
Summary
The local run passed — but that only proves consistency with itself, not with CI. Ran it in CI, found a real ~2% pixel diff from font-rendering differences, fixed the baseline using CI's own actual output, and re-verified for real.
| Check | Status |
|---|---|
| Local run (Docker) | Passed — but not proof of CI match |
| CI run (Visual Regression job) | Failed — 2/2, ~2% pixel diff |
| After baseline fix | Genuinely verified |
{
"checksRun": [
{ "command": "npm run test:visual (local, Docker)", "status": "passed", "summary": "2/2 — only proves consistency with itself, not with CI" },
{ "command": "npm run test:visual (CI)", "status": "failed", "summary": "2/2 failed — font rendering differs between local Docker and CI's runner" }
],
"overallStatus": "failed",
"residualRisks": ["Baseline was generated in an environment that didn't match CI closely enough — documented as a repeatable correction procedure."]
}
Next steps
- Correct the baseline using CI's own actual-output images, not a locally-approximated one
- Re-run and confirm overallStatus: "verified" before calling this done
Connects securely to your tools. The creator never sees your data.
About this skill
The gap between "I wrote code that should do X" and "I ran the real thing and confirmed it does X" is where most false "it's done" claims live. This skill closes that gap: it reviews the actual diff for anything unintended, runs the narrowest relevant tests first then the required broader checks — using your repository's own real commands, never an assumption that something "probably still passes."
It treats a local pass and a CI pass as different claims — if CI's own result disagrees with a local run, the CI failure is what gets reported, not smoothed over as a fluke. Every check gets an honest status: passed, failed, or not-run (and "not-run" is never silently treated as "fine"). If something fails, it either gets fixed and re-verified, or reported honestly — this skill will never weaken or delete a meaningful test just to get a green result.
Part of the Ticket-to-Shipped workflow: 10 skills covering the full requirement-to-shipped-code lifecycle. Works standalone too.
Find more skills like this on the Claude Skills Marketplace, the Codex Skills Marketplace and the Gemini Skills Marketplace, or browse Claude Skills for Testing & Debugging.
Only as thorough as the toolchain available in your environment — if a required tool isn't installed (e.g. a browser binary for E2E), it reports "incomplete" honestly rather than skipping silently. Doesn't fix failing code itself beyond what's needed for a real pass. Assumes a prior planning step already told it which checks are relevant — without that, it may not know your stack's full check list.
How to use
- 1
Connect your agent to the Agensi MCP
Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup
- 2
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 3
Ask for what you need
Your agent loads this skill and uses it on your task. It always asks before buying a skill.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill goes through the trust checks below.
Security scanned
Checked by our 8-point scan before listing
Fresh listing
Recently published to Agensi
Money-back guarantee
Doesn't do what its page promises? You get your money back.
Trust & safety
Security scan
Security scan passed 18 days ago
- Software
- Your project's test and build tools
GitPlaywright
Node.js
Whatever your stack uses to test, build, lint and scan, such as Maven, npm or Playwright.
Permissions
Never commits anything — build artifacts (target/, coverage/, node_modules/) are ephemeral outputs of running checks, not tracked changes.