loop-design-check

Design a goal-oriented agent loop, and review it for the ways loops go wrong — spinning and burning tokens, Goodhart-gaming the verifier, or running a wrong answer to completion. Two actions: (1) WRITE a loop — gate whether to build it, define a machine-decidable goal, pick the loop type, pick a skeleton; (2) REVIEW a loop — run it past five failure modes plus decidability, boundaries, fallback, judge independence, and keep-judgment-with-the-human red lines. Use when designing an autonomous agent loop, or when you already have one and worry it will spin, cheat, or run a wrong answer to the end. Complements the mechanism-layer loop skills (autonomous-loops, continuous-agent-loop) by covering the judgment layer they don't. 中文触发：写 loop、设计 loop、做一个 loop、检查 loop 对不对、loop 体检、loop 会不会跑飞、可判定目标、五个崩法、plan build judge。English triggers: design an agent loop, write a loop, check a loop, loop review, prevent a runaway loop, goal-oriented loop, decidable goal, plan/build/judge.

Quality

85%

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

Quality

Content

70%

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

The skill body is well-structured with clear sequenced workflows, explicit validation checkpoints, and a strong review checklist, but it is monolithic with no bundle references and leans conceptual rather than providing executable templates. Tightening the repeated motivational passages would improve token efficiency.

Suggestions

Trim the repeated premise/thermostat/wrong-output warnings (Premise, Red-line premise, and One-line close restate the same idea) to reduce tokens without losing the point.

Add a small references/ or scripts/ file (e.g., a ready-to-use goal-definition or review-checklist template) so the detailed tables/worked example can be one level deep instead of inline.

Provide one concrete fill-in-the-blank template (e.g., a sample decidable-goal statement + boundary + retry-cap config) to lift actionability from checklist to copy-paste ready.

Dimension	Reasoning	Score
Conciseness	The body is dense and largely avoids re-explaining known concepts, but the motivational premise, repeated thermostat/wrong-output warnings, and one-line close restate the same insight multiple times; it could be tightened without losing substance.	2 / 3
Actionability	It gives concrete frameworks (4-condition gate, five-point goal framework, failure-mode table) but stays at the conceptual/checklist level — no executable commands, scripts, or copy-paste examples, which fits an instruction-only design skill but leaves guidance short of fully concrete templates.	2 / 3
Workflow Clarity	Action 1 lays out an explicitly numbered 5-step (Step 0–Step 5) sequence with a self-check checkpoint, and Action 2 is a table-based review checklist with a worked example and explicit red lines/validation gates (retry cap, human flips last switch, escalate after 3 fails).	3 / 3
Progressive Disclosure	No references/, scripts/, or assets/ bundle exists, and the body is a single ~140-line monolithic document with no one-level-deep references to split out the detailed tables or worked example; it is well-organized with section headers but content that could be separate remains inline.	2 / 3
	Total	9 / 12 Passed

Description

100%

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

The description is strong across all dimensions: it lists concrete write/review actions, provides bilingual natural trigger terms, states both what and when explicitly, and clearly distinguishes itself from sibling mechanism-layer skills. It is long but the detail is substance rather than fluff.

Dimension	Reasoning	Score
Specificity	Names multiple concrete actions — 'WRITE a loop — gate whether to build it, define a machine-decidable goal, pick the loop type, pick a skeleton' and 'REVIEW a loop — run it past five failure modes plus decidability, boundaries, fallback, judge independence, and keep-judgment-with-the-human red lines' — matching the 'lists multiple specific concrete actions' anchor.	3 / 3
Completeness	Explicitly answers 'what' (two actions: WRITE and REVIEW with their steps) and 'when' ('Use when designing an autonomous agent loop, or when you already have one and worry it will spin, cheat, or run a wrong answer to the end').	3 / 3
Trigger Term Quality	Includes both Chinese ('写 loop、设计 loop、做一个 loop、检查 loop 对不对、loop 体检') and English ('design an agent loop, write a loop, check a loop, loop review, prevent a runaway loop, decidable goal, plan/build/judge') natural trigger phrases a user would actually say.	3 / 3
Distinctiveness Conflict Risk	Carves a clear niche ('covers the judgment layer' that 'autonomous-loops / continuous-agent-loop' do not) and explicitly demarcates scope, making it unlikely to trigger for the wrong skill.	3 / 3
	Total	12 / 12 Passed

Validation

93%

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 — 15 / 16 Passed

Validation for skill structure

Criteria	Description	Result
metadata_version	'metadata.version' is missing	Warning

	Total	15 / 16 Passed

Repository: affaan-m/ECC
Path: skills/loop-design-check/SKILL.md
Commit: 0c1d7be

Reviewed: 2 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.