auto-review-loop

Autonomous multi-round research review loop. In Copilot CLI it defaults to the native complementary rubber-duck subagent with host-event model evidence; elsewhere it uses Codex, while explicit external reviewer overrides remain available. Implements fixes and re-reviews until a policy-approved positive assessment or max rounds is reached.

Quality

64%

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 ./skills/auto-review-loop/SKILL.md

Quality

Content

70%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.

The body is exceptionally actionable and the multi-round workflow is clearly sequenced with strong validation gates. Its main weaknesses are verbosity from repeated boilerplate and poor progressive disclosure — substantial inlined content that should be split into referenced files.

Suggestions

Factor the repeated SCOPE LIMITS block into a single shared-references file and reference it once instead of pasting it three times.

Move the full reviewer prompt templates and the review_gate.py bash block into referenced script/reference files, leaving the SKILL.md body as an overview.

Extract the REVIEW_STATE.json and ACQUITTAL_LOG.jsonl schemas into a dedicated reference file to reduce inline bulk.

Dimension	Reasoning	Score
Conciseness	The ~1130-line body is noticeably to severely verbose: the SCOPE LIMITS block is duplicated verbatim three times, and identity-assurance / acquittal prose is repeated across sections, inflating the token budget without adding capability.	2 / 5
Actionability	Guidance is highly concrete and executable — exact MCP calls with config, copy-paste bash (codex exec heredocs, copilot --agent invocations, the review_gate.py GATE_ARGS array), exact JSON schemas and run-scoped filenames — covering the common backend cases.	5 / 5
Workflow Clarity	The Initialization → Loop (Step -1/0, Phases A–E) → Termination sequence is explicit, with validation checkpoints (Phase B.5.1 stop-gate, fail-closed REVIEW_UNAVAILABLE, evidence revalidation), feedback loops (review→fix→re-review), and a test checklist.	5 / 5
Progressive Disclosure	External references (shared-references/.md, .aris/tools/) are clearly signaled and one level deep, but no bundle files exist and large blocks that belong in separate files — full prompt templates, the gate bash script, the state/acquittal JSON schemas — are inlined in the monolithic body.	3 / 5
	Total	15 / 20 Passed

Description

58%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.

The description conveys a clear, specific purpose but omits explicit 'when to use' trigger guidance and leans on backend-implementation jargon over natural user language. It is reasonably distinctive but would benefit from a concrete 'Use when...' clause.

Suggestions

Add an explicit 'Use when ...' clause naming natural trigger phrases (e.g., 'Use when you want an independent reviewer to iteratively critique and improve research work').

Lead with the user-facing capability (autonomous review→fix→re-review loop) before the backend-routing details so trigger terms surface first.

Include common synonyms users might say ('review my experiments', 'critique my paper draft', 'adversarial review loop') to improve trigger term coverage.

Dimension	Reasoning	Score
Specificity	Names several concrete actions — 'multi-round research review loop', 'implements fixes and re-reviews', 'until a policy-approved positive assessment or max rounds is reached' — though they are entangled with backend-implementation jargon rather than presented as a clean capability list.	4 / 5
Completeness	It clearly states what the skill does, but provides no 'Use when...' clause or equivalent explicit trigger guidance, which per the rubric caps completeness at 3.	3 / 5
Trigger Term Quality	Relevant keywords like 'research review loop' and 'review' appear, but the description leans on implementation terms (Copilot CLI, rubber-duck subagent, Codex) rather than natural phrases a user would say, and lacks synonyms or variations.	3 / 5
Distinctiveness Conflict Risk	The 'autonomous multi-round research review loop' niche is mostly distinct with low conflict risk, though it could overlap with adjacent review/critique skills and lacks explicit distinguishing triggers.	4 / 5
	Total	14 / 20 Passed

Validation

75%

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 — 12 / 16 Passed

Validation for skill structure

Criteria	Description	Result
skill_md_line_count	SKILL.md is long (1138 lines); consider splitting into references/ and linking	Warning
allowed_tools_field	'allowed-tools' contains unusual tool name(s)	Warning
frontmatter_unknown_keys	Unknown frontmatter key(s) found; consider removing or moving to metadata	Warning
relative_links	Relative link issues: 4 suspicious	Warning

	Total	12 / 16 Passed

Repository: wanshuiyin/Auto-claude-code-research-in-sleep
Path: skills/auto-review-loop/SKILL.md
Commit: 0472e53

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.