Skip to content
Back to skills

Flow Verify

ASecurity

Adversarially verify an implementation against its objective, acceptance criteria, definition of done, and prohibitions using fresh evidence. Use for completion gates, release readiness, or independent verification after implementation and review.

  • 3 stars
  • 0 votes
  • 0 copies
  • 2 views
  • Added September 9, 2026
testingrustgo

Security analysis

A100/100

Pro scans all 2 files and shows the line behind each finding

Scanned September 9, 2026

npx -y skills add pwdev-solucoes/pwdev-claude-marketplace --skill flow-verify --agent claude-code

Installs into .claude/skills of the current project.

Are you the author of Flow Verify?

Add the live security badge to your README. It updates with every re-scan.

Security grade badge for Flow Verify
[![Security: A — Skills Directory](https://www.skillsdirectory.com/api/skills/pwdev-solucoes-flow-verify/badge)](https://www.skillsdirectory.com/skills/pwdev-solucoes-flow-verify)

More formats (shields.io, HTML) on the badges page. Keep it an A: scan every change in CI with Pro.

Download with Pro
SKILL.md
---
name: flow-verify
description: Adversarially verify an implementation against its objective, acceptance criteria, definition of done, and prohibitions using fresh evidence. Use for completion gates, release readiness, or independent verification after implementation and review.
---

# Verify completion adversarially

Read [workflow](../../references/workflow.md), [artifacts](../../references/artifacts.md), and [safety](../../references/safety.md) before verification.

The goal is to refute completion, not confirm the implementation narrative. Treat summaries and prior test claims as untrusted until reproduced.

## Inputs

Resolve, in order:

1. objective;
2. acceptance criteria;
3. definition of done;
4. prohibitions and safety constraints;
5. changed files and implementation evidence;
6. prior review findings.

If no verifiable contract exists, return `BLOCKED` and list the missing truths. Do not invent acceptance criteria after implementation.

## Procedure

1. Convert every contract statement into an independently testable truth.
2. For each truth, choose the command, inspection, or counterexample most likely to disprove it.
3. Run fresh verification and record the exact result; do not rely on earlier output.
4. Inspect the final diff for scope violations and prohibited behavior.
5. Reconcile unresolved review findings with the truth list.
6. Calculate passed truths and identify critical failures.
7. When rejected, write bounded correction tasks only if artifact persistence is authorized.

## Verdicts

- `APPROVED`: every truth passes and no prohibition is violated.
- `WITH_CAVEATS`: all critical truths pass; only explicitly low-impact caveats remain.
- `REJECTED`: any critical truth fails, a prohibition is violated, or required evidence contradicts completion.
- `BLOCKED`: verification cannot run or the contract is absent.

Never weaken a failing criterion to obtain approval.

## Output

Return a truth table with `Truth`, `Refutation attempt`, `Evidence`, and `Result`, followed by the verdict and correction paths. Persist the verification report only when the user requested it or an active Flow phase already authorizes artifacts.

Files in this skill

  • SKILL.md2.1 KB
  • agents/openai.yaml205 B

Attribution

Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.

Comments

Loading comments…