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

55%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide

Securityby

Medium

Suggest reviewing before use

Fix and improve this skill with Tessl

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

Quality

Content

56%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 skill is highly actionable with a well-sequenced, validation-guarded workflow, but it is dramatically over-long with repeated blocks and inlined detail that should live in separate reference files. Progressive disclosure is mediocre because the bundle directories are empty.

Suggestions

De-duplicate the SCOPE-LIMITS block (pasted five times) into a single shared-references file referenced once per prompt, cutting hundreds of lines.

Move the large reviewer-routing/Copilot contract prose into a referenced file (it already cites reviewer-routing.md) and keep only the routing decision table inline.

Populate ./references/ with the cited shared-references files (reviewer-routing.md, review-tracing.md, external-cadence.md) so the progressive-disclosure structure is real rather than just signaled.

Dimension	Reasoning	Score
Conciseness	The body is severely long (~1100 lines) with the identical SCOPE-LIMITS block pasted five times and heavily duplicated reviewer-routing prose across Constants, Calling Convention, and Phase A; this is noticeably verbose with large padded sections.	2 / 5
Actionability	Provides concrete executable commands (codex exec, copilot --agent invocations) and complete MCP call templates with configs; only minor gaps remain as runtime-filled placeholder paths.	4 / 5
Workflow Clarity	Multi-step phases are clearly sequenced with explicit validation checkpoints (review_gate.py, evidence verify, fail-closed REVIEW_UNAVAILABLE) and feedback loops for the destructive batch loop; a few minor validation gaps keep it below 5.	4 / 5
Progressive Disclosure	Section structure exists and points to shared-references/* files, but no bundle files exist in the skill directory and most detail is inlined rather than split out; references are present but not consistently signaled one level deep.	3 / 5
	Total	13 / 20 Passed

Description

53%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 niche and a couple of concrete actions but is missing an explicit "Use when..." trigger clause, which caps completeness. It is reasonably distinctive but not optimally specific or trigger-rich.

Suggestions

Add an explicit 'Use when ...' clause naming the natural trigger phrases (e.g., 'Use when the user wants an independent multi-round critique of research work or asks to review/improve a paper draft').

List 2-3 more concrete actions in the description (e.g., 'scores work 1-10, ranks weaknesses, runs adversarial debate') rather than leaving specifics only in the body.

Include common synonyms users would say ('peer review', 'paper critique', 'review my draft') to improve trigger-term coverage.

Dimension	Reasoning	Score
Specificity	Names the domain (autonomous research review loop) and one or two concrete actions ("implements fixes and re-reviews"), but the bulk of concrete capability lives in the body rather than the description, so coverage is not comprehensive.	3 / 5
Completeness	States a clear "what" but has no explicit "Use when..." clause; the "when" is only weakly implied by "Autonomous multi-round research review loop," which per the rubric caps completeness at 3.	3 / 5
Trigger Term Quality	Includes relevant terms ("research review", "review loop", "positive assessment") but omits the natural phrases a user would actually say ("review my paper", "critique this work") and common synonyms.	3 / 5
Distinctiveness Conflict Risk	Clearly niched to multi-round cross-model research review with reviewer-backend semantics; only minor overlap risk with a generic peer-review skill, so it sits above the midpoint but not at the fully-distinct anchor.	4 / 5
	Total	13 / 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 (1118 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: f4f20f9

Reviewed: 1 day 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.