iterate-until-verified

Apply a prompt-agnostic execution and verification loop to any substantial task while preserving the original request. Use when the user asks to fan out work, use subagents or independent reviewers, loop until done, benchmark against references, apply a harsh critic, compare candidates blind, improve an existing prompt with verification, or continue until explicit quality gates pass.

Quality

80%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide

Securityby

Passed

No findings from the security scan

Fix and improve this skill with Tessl

tessl review fix ./agent-skills/codex/iterate-until-verified/SKILL.md

Quality

Content

81%Weight 40%Scale 1-5

Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.

A well-crafted process skill: tightly written, honestly structured around gates and evidence, with an explicit feedback loop and stop conditions. Its only real gaps are minor - a few self-evident principles that could be trimmed and one or two sections (the compose template, verification-surface menus) that would benefit from being split into reference files for progressive disclosure.

Suggestions

Trim principle statements Claude already knows (e.g., 'Do not let an implementer be the sole approver of its own work' and the blind-comparison hygiene bullets) to tighten conciseness.

Move the Compose-mode prompt template into a references/ file and point to it from SKILL.md to improve progressive disclosure as the skill grows.

Add one short worked example of a filled-in acceptance matrix (gate, method, pass condition, evidence) to make the gate-conversion step immediately copy-paste ready.

Dimension	Reasoning	Score
Conciseness	The body is lean and imperative with essentially no padding or explanation of concepts Claude already knows. A few general principle statements (e.g., 'Do not let an implementer be the sole approver of its own work') restate widely known ideas and could be trimmed.	4 / 5
Actionability	Concrete, executable guidance throughout: a fill-in acceptance-matrix table, a 7-step numbered loop, and a copy-paste Compose-mode prompt template. Verification surfaces are listed as option menus rather than specific commands or worked examples, leaving minor gaps.	4 / 5
Workflow Clarity	Sections 1-7 form a clearly sequenced process with explicit validation checkpoints ('Record pass, fail, or blocked with evidence', 'Re-run the failed gate and any affected regression gates'), a route-failure-to-owner feedback loop, and a completion checklist at the end - matching the top anchor.	5 / 5
Progressive Disclosure	A single, well-sectioned file with clear headers and no nested or broken references (no bundle files exist). At ~150 lines it exceeds the simple-skill exception, and content like the Compose-mode template or verification-surface menus could arguably live in reference files, but overall placement is appropriate.	4 / 5
	Total	17 / 20 Passed

Description

78%Weight 40%Scale 1-5

Based on the skill's description, can an agent find and select it at the right time? Clear, specific descriptions lead to better discovery.

A strong description with an explicit and richly phrased 'Use when' clause covering eight distinct trigger scenarios. Its main weaknesses are the somewhat abstract 'what' statement and a broad subject scope ('any substantial task') that slightly raises overlap risk with general execution skills.

Dimension	Reasoning	Score
Specificity	The description names its domain ('execution and verification loop') and one concrete action ('preserving the original request'), but 'apply a prompt-agnostic... loop' is abstract and does not enumerate several concrete capabilities the way top anchors expect.	3 / 5
Completeness	It explicitly answers both what ('Apply a prompt-agnostic execution and verification loop... while preserving the original request') and when ('Use when the user asks to fan out work... or continue until explicit quality gates pass') with eight concrete trigger phrases, matching the top anchor's pattern.	5 / 5
Trigger Term Quality	Natural trigger phrases are well covered ('fan out work', 'use subagents', 'loop until done', 'harsh critic', 'compare candidates blind', 'quality gates'), matching phrases users would actually say. A few natural variants ('verify', 'iterate', 'adversarial review') are missing, keeping it below the comprehensive anchor.	4 / 5
Distinctiveness Conflict Risk	The trigger phrases (subagents, blind comparison, harsh critic, quality gates) form a distinct niche unlikely to fire for other skills. However, 'any substantial task' is a broad subject that overlaps generic execution and review skills, so conflict risk is minor but not minimal.	4 / 5
	Total	16 / 20 Passed

Validation

100%

Warnings & errors only

Checks the skill against the spec for correct structure and formatting. All validation checks must pass before discovery and implementation can be scored.

Validation — 16 / 16 Passed

Validation for skill structure

No warnings or errors.

Repository: MengTo/Skills
Path: agent-skills/codex/iterate-until-verified/SKILL.md
Commit: 798db0a

Reviewed: 5 days ago

Table of Contents

Discovery Implementation Validation

Is this your skill?

If you maintain this skill, you can claim it as your own. Once claimed, you can manage eval scenarios, bundle related skills, attach documentation or rules, and ensure cross-agent compatibility.