---
title: "verification-loop"
description: "Enforce a claim-to-proof workflow: define the evidence, run the right verification, and distinguish verified results from assumptions before declaring work complete."
canonical: https://agentpluginsdirectory.com/plugins/verification-loop
last-updated: 2026-10-02
---

# verification-loop
Enforce a claim-to-proof workflow: define the evidence, run the right verification, and distinguish verified results from assumptions before declaring work complete.
- Slug: verification-loop
- Publisher: Hylouis233
- Repository: https://github.com/hetaoBackend/MiniMax-Code-Plugins
- Manifest: plugins/Hylouis233/verification-loop/plugin.json
- Version: 0.1.0
- License: Apache-2.0
- Category (editorial): code-review
- Skills: 1 (verification-loop)
- MCP servers: 0
- Stars: 12
- Repository created: 2026-08-14
- Repository last pushed: 2026-09-07
- Publisher type: User
- Listing: https://agentpluginsdirectory.com/plugins/verification-loop
- Schema: https://agent-plugins.org/schemas/1.0.0/plugin.schema.json

## What verification-loop does, in the publisher's words

Agents routinely declare work "done" without evidence: the fix was written but never run, the test suite was never invoked, or a claim of "works now" rests on confidence rather than output. Downstream, a human discovers the gap only when something breaks.

This Plugin installs one Skill that enforces a claim-to-proof loop: state the claim, decide what evidence would prove it, run the smallest meaningful verification, read the actual result, and report verified / not verified / could-not-verify explicitly.

Expected result: the agent states each claim it is about to make ("page 2 returns 11-20"), runs the smallest check that proves it (the endpoint or the targeted test), quotes the real output, and labels anything it could not check as not verified.

From the project README, punctuation lightly normalized. Full text: https://raw.githubusercontent.com/hetaoBackend/MiniMax-Code-Plugins/HEAD/plugins/Hylouis233/verification-loop/README.md

## Skills

- verification-loop: Enforce a claim-to-proof workflow before declaring work complete. Use this Skill when a task involves implementation, debugging, review, or any other work that could be claimed complete without concrete evidence.

Descriptions come from the frontmatter of each SKILL.md, punctuation lightly normalized.