do-and-judge

Execute a task with sub-agent implementation and LLM-as-a-judge verification with automatic retry loop

Quality

60%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide

Securityby

Critical

Do not install without reviewing

Fix and improve this skill with Tessl

tessl review fix ./plugins/sadd/skills/do-and-judge/SKILL.md

Quality

Content

62%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 highly actionable and its six-phase workflow is among the clearest and best-validated one could ask for, but it pays for that with significant verbosity and a monolithic structure that inlines material (especially the five worked examples) that belongs in separate reference files.

Suggestions

Move the five worked examples (Examples 1-5, which re-render full prompt templates verbatim) into a separate EXAMPLES.md referenced once from the Process section, cutting several hundred duplicated tokens.

Extract the Cross-Provider Equivalence table and the full prompt templates into reference files, keeping only the decision rules and dispatch signatures inline in SKILL.md.

Trim editorializing passages (e.g., 'Picking the model is the single highest-leverage decision you make…') to directive statements that preserve the rule without the persuasion.

Dimension	Reasoning	Score
Conciseness	The 1241-line body is noticeably verbose: the full meta-judge and implementation prompt templates are re-rendered verbatim inside Examples 1, 4, and 5, and several editorializing passages ('single highest-leverage decision') could be trimmed, fitting the score-2 anchor of several padded sections without sinking to the 'explains concepts Claude already knows' level of 1.	2 / 5
Actionability	Concrete copy-paste-ready prompt templates, exact Task-tool dispatch specs, decision-logic pseudocode, and YAML report formats give mostly executable guidance, with only template placeholders ({task}, {score}) keeping it just below fully copy-paste-ready at 5.	4 / 5
Workflow Clarity	Six phases are clearly sequenced with explicit validation checkpoints (judge verification, score thresholds), feedback loops (judge feedback → retry → re-judge), and a self-critique checklist, matching the score-5 anchor; the judge-verification phase supplies the validation that the destructive/batch cap would otherwise require.	5 / 5
Progressive Disclosure	Section headers give it real structure, but the skill is a monolithic 1241 lines with five long worked examples and the cross-provider equivalence table inlined that clearly belong in separate reference files, and no bundle files exist or are referenced, fitting the score-3 anchor of content that should be separate being inline.	3 / 5
	Total	14 / 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 clearly conveys what the skill does through several concrete mechanisms, but lacks any 'when to use it' trigger guidance and leans on technical jargon over natural user phrasing. It is distinct but not as discoverable as it could be.

Suggestions

Add an explicit 'Use when…' clause with concrete trigger phrases users would naturally say (e.g., 'Use when you need a task implemented and independently verified, or when you want automatic retries with judge feedback').

Soften the technical jargon ('LLM-as-a-judge verification', 'sub-agent implementation') or pair each with a plain-language synonym so the description reads as something a user would actually say.

Name the common cases covered (code, documentation, configuration) to lift specificity toward comprehensive coverage.

Dimension	Reasoning	Score
Specificity	The description names the domain and several concrete actions — 'sub-agent implementation', 'LLM-as-a-judge verification', 'automatic retry loop' — matching the score-4 anchor of several specific actions with minor coverage gaps rather than the comprehensive list needed for 5.	4 / 5
Completeness	It clearly answers 'what' but provides no 'Use when…' or equivalent trigger guidance, so per the rubric guideline completeness is capped at 3.	3 / 5
Trigger Term Quality	Terms like 'sub-agent implementation' and 'LLM-as-a-judge verification' are relevant but technical rather than the natural phrases a user would say, and common synonyms/variations are missing, fitting the score-3 anchor.	3 / 5
Distinctiveness Conflict Risk	The combination of sub-agent dispatch plus judge verification plus retry loop is a fairly distinct niche with only minor overlap risk against generic task-execution skills, matching the score-4 anchor.	4 / 5
	Total	14 / 20 Passed

Validation

87%

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 — 14 / 16 Passed

Validation for skill structure

Criteria	Description	Result
skill_md_line_count	SKILL.md is long (1242 lines); consider splitting into references/ and linking	Warning
frontmatter_unknown_keys	Unknown frontmatter key(s) found; consider removing or moving to metadata	Warning

	Total	14 / 16 Passed

Repository: NeoLabHQ/context-engineering-kit
Path: plugins/sadd/skills/do-and-judge/SKILL.md
Commit: 8539779

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.