Skip to content
Get packs

Iterate Until Verified

Makes the agent run tasks through a verify-and-revise loop with explicit pass/fail quality gates

Source Download skill
Repository
MengTo/Skills
Stars
7,187
Licence
MIT
Files
3 files

What it does

Once installed, the agent treats substantial tasks as a loop: it turns vague quality words like 'perfect' or 'production-ready' into pass/fail checks, splits work into parts, and has a separate reviewer check the output against evidence before declaring it done. It can also rewrite a user's prompt into a reusable prompt wrapper that keeps the original task unchanged. The skill itself is only instructions; it adds process, not new tools.

Great for

  • People who want an agent to double-check its own substantial work before reporting back
  • Teams building prompts or evaluation workflows that need explicit acceptance criteria
  • Users asking for multi-step work that benefits from independent review

Skip it if

  • Quick, simple questions, where the loop adds overhead without value
  • Situations needing guaranteed verification: the skill relies on the agent actually running checks, which it can misreport or skip
  • Tasks where the agent lacks permission to use subagents or tools; the skill then falls back to sequential work

Try asking

  • “Iterate until verified: build this landing page and check it renders correctly at mobile and desktop sizes”
  • “Compose a better prompt for this research task, keeping my original request unchanged”
  • “Use independent reviewers to check this report's facts against the sources before you finish”
  • “Apply a harsh critic to this draft and loop until it passes the gates we define”
Step-by-step guide: how to use Iterate Until VerifiedInstall, first run, examples, tips and FAQs

Install in 3 steps

  1. 1

    Download the skill

    The ZIP contains the 3 original files, the MIT licence and an attribution note.

    iterate-until-verified.zip
  2. 2

    Add it to your agent

    Claude Code (all projects)

    mkdir -p ~/.claude/skills && unzip iterate-until-verified.zip -d ~/.claude/skills/

    Claude Code / Agent SDK (this project only)

    mkdir -p .claude/skills && unzip iterate-until-verified.zip -d .claude/skills/

    Claude apps (web/desktop): Settings → Capabilities → Skills → Upload skill, and choose the ZIP.

  3. 3

    Ask for the task

    No special command needed — the agent reads the skill's description and loads it when your request matches. Try one of the prompts above.

Get better results

  • State your definition of done up front so the agent can turn it into gates
  • Ask for compose mode explicitly if you want a reusable prompt instead of the finished work
  • Check the evidence the agent reports, since the skill asks it to cite evidence but cannot force it to run real tests

What's inside

  • agents/openai.yaml
  • REFERENCES.md
  • SKILL.md

Redistributed unmodified from MengTo/Skills under the MIT licence. The assessment was drafted with AI from the skill's files on 2026-10-29. Skills can instruct your agent to run code — read SKILL.md and any scripts before installing.