ds-optimize

Use when an algorithm-first quest should manage candidate briefs, optimization frontier, branch promotion, or fusion-aware search instead of the paper-oriented default loop.

Quality

63%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide

Securityby

Low

Low-risk findings worth noting

Fix and improve this skill with Tessl

tessl review fix ./skills/ds-optimize/SKILL.md

Quality

Content

58%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 delivers exceptionally clear, well-validated workflows and concrete actionable calls, but its quality is undermined by extreme length, heavy internal duplication, and a deliberate anti-progressive-disclosure decision to inline and repeat reference material.

Suggestions

De-duplicate the body: the "Prompt patterns worth preserving", brief-shaping sequence, candidate-brief field lists, artifact call lists, and first-call/stage-start sequences each appear multiple times — keep one canonical statement and reference it.

Restore progressive disclosure by moving the templates and playbooks in the "Integrated reference appendix" into actual separate reference files under ./references/ and linking to them one level deep, instead of inlining them twice.

Provide one fully worked, end-to-end example of an artifact.submit_idea call with a complete payload to lift actionability from mostly-executable to copy-paste ready.

Dimension	Reasoning	Score
Conciseness	At ~1650 lines the body is noticeably verbose with severe structural duplication: the "Prompt patterns worth preserving" section, brief-shaping sequence, candidate-brief fields, artifact calls, and mandatory first-call sequence each recur 2-4 times, and the "Integrated reference appendix" re-inlines ~540 lines already covered inline.	2 / 5
Actionability	Highly actionable with concrete MCP calls and arguments (artifact.submit_idea(mode='create', submission_mode='candidate', ...), memory.list_recent(scope='quest', limit=5), bash_exec(mode='detach', ...)), named control files, numeric bounds (2-3 candidates, 20-minute fast/slow-check, success_patience >= 2), status enums, and an example CANDIDATE_BOARD table; minor gaps remain from ellipsis payload placeholders and the absence of one fully filled end-to-end call.	4 / 5
Workflow Clarity	Multi-step workflows are explicitly sequenced (7-step Required MCP-driven workflow, 7-step Recommended workflow, mandatory first-call sequence) with explicit validation checkpoints (smoke-then-full-eval gating, read logs before retry, post-fix bounded checks), feedback loops (debug/archive on failure, plateau route-change, family-shift trigger), and a checkbox checklist template.	5 / 5
Progressive Disclosure	No bundle files exist and the skill is a single ~1650-line file that deliberately inlines former separate reference files ("This appendix inlines the former optimize/references/*.md material so the skill remains self-contained"), duplicating content that clearly belongs in separate files despite headers providing some structure.	2 / 5
	Total	13 / 20 Passed

Description

67%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 domain-specific, reasonably concrete description with an explicit "Use when" trigger and a clear niche discriminator, weakened by internal jargon over natural user phrasings and a single umbrella verb covering its actions.

Suggestions

Lead with plain trigger verbs a user would actually say (e.g. "Optimize an algorithm-first quest: generate and rank candidate briefs, promote branches, run fusion-aware search") rather than the umbrella "manage".

Add natural synonyms users might say ("optimize", "improve the result", "search for better approaches") alongside the internal terms like "optimization frontier" and "fusion-aware search".

Separate the "what" (actions) from the "when" (trigger conditions) into distinct clauses so the trigger reads as user-facing rather than fused with the action list.

Dimension	Reasoning	Score
Specificity	Names the optimization domain and several concrete action areas ("manage candidate briefs, optimization frontier, branch promotion, or fusion-aware search"), but joins them under a single umbrella verb "manage" rather than distinct concrete verbs, a minor gap versus the multi-verb score-5 anchor.	4 / 5
Completeness	An explicit "Use when" clause is present and names a concrete discriminator ("an algorithm-first quest ... instead of the paper-oriented default loop"), answering both what and when, but the trigger is fused into the action clause and is system-state-facing rather than user-mention-facing like the score-5 anchor.	4 / 5
Trigger Term Quality	Relevant domain keywords are present ("algorithm-first quest", "optimization frontier", "fusion-aware search", "branch promotion") but they lean toward internal system jargon; plain user-facing synonyms like "optimize", "improve", or "search for better approaches" are missing.	3 / 5
Distinctiveness Conflict Risk	The "instead of the paper-oriented default loop" carve-out gives it a clear niche, but it risks minor overlap with the closely related experiment stage that also manages runs and candidates within the same ecosystem.	4 / 5
	Total	15 / 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 (1650 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: OpenLAIR/dr-claw
Path: skills/ds-optimize/SKILL.md
Commit: 4b35756

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.