ce-optimize

Optimize a named target with a measured loop: attribute a workload's cost, or score variants and keep winners. Use when a working system's metric should move and the winning change is not already known. Use ce-debug when the job is diagnosis; use ce-work when the change is already known.

Quality

80%

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

The canonical home for this skill is ce-optimize in EveryInc/compound-engineering-plugin

Quality

Content

81%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.

A well-structured orchestrator skill: a compact phased workflow with explicit gates, checkpoints, and stopping criteria, backed by a real one-level reference bundle. The only weaknesses are minor — conduct boilerplate in the body and slightly deep reference chaining.

Dimension	Reasoning	Score
Conciseness	Lean, imperative, and free of concept explanation; every section carries operational content and detail is pushed to references. Not 5 because the reporting/interaction-policy paragraphs ("Routine preparation and phase or batch transitions need no separate announcement") are agent-conduct boilerplate not specific to optimization and could be trimmed.	4 / 5
Actionability	Concrete guidance throughout: named artifacts ("scripts/decide.mjs", "optimize/<spec-name>" branch), a literal question to ask, exact config key names ("scope.mutable", "max_total_cost_usd"), and per-phase "Read references/X" directives. Not 5 because executable commands live in the references rather than the body, so the body alone is not copy-paste runnable — acceptable for an orchestrator skill.	4 / 5
Workflow Clarity	Four phases in explicit order with numbered checkpoints (CP-0 through CP-5), hard gates (clean-tree gate, user approval gate, dependency pre-approval), seven enumerated stopping criteria, resume semantics, and an error path ("If the run instead stopped at a check it could not pass, say what blocked it"). Validation and feedback loops are explicit for the destructive/batch operations, meeting the top anchor.	5 / 5
Progressive Disclosure	Each phase names the reference it cannot start without, all referenced files (references/spec.md, measurement.md, loop.md, wrap-up.md, persistence.md, scripts/decide.mjs) exist, and the body is a genuine overview. Not 5 because references chain one level further (loop.md → experiment-prompt-template.md, spec.md → example specs), and there is no index of the full bundle, so navigation depends on reading the phase text.	4 / 5
	Total	17 / 20 Passed

Description

80%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 strong description: third-person voice, concrete actions, and an explicit trigger clause with sibling-skill routing. The only gap is synonym coverage for natural trigger terms.

Suggestions

Add one or two natural synonyms users would say, e.g. "Use when asked to speed up, reduce cost of, or otherwise improve a measured target".

Consider naming the artifact users start from ("an optimization spec YAML") in the trigger clause, since users often invoke by describing the file they have.

Dimension	Reasoning	Score
Specificity	Lists several concrete actions — "attribute a workload's cost", "score variants and keep winners", "optimize a named target with a measured loop" — plus routing actions for sibling skills. Not a 5 because "measured loop" is abstract phrasing and coverage of what the loop actually does is deferred to the body.	4 / 5
Completeness	Explicitly answers what ("attribute a workload's cost, or score variants and keep winners") and when ("Use when a working system's metric should move and the winning change is not already known"), with concrete trigger conditions plus counter-triggers for ce-debug/ce-work. Both halves are explicit and specific.	5 / 5
Trigger Term Quality	"Optimize", "metric should move", and "diagnosis" are natural user phrases, but common synonyms users would say are missing ("speed up", "make it faster", "improve performance", "reduce cost"). Fits anchor 3 (relevant keywords, missing common variations) better than anchor 4, which expects fuller synonym coverage.	3 / 5
Distinctiveness Conflict Risk	Occupies a clear niche (measured optimization loop with cost attribution and variant scoring) and actively disambiguates from sibling skills ("Use ce-debug when the job is diagnosis; use ce-work when the change is already known"), giving minimal conflict risk.	5 / 5
	Total	17 / 20 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
frontmatter_unknown_keys	Unknown frontmatter key(s) found; consider removing or moving to metadata	Warning

	Total	15 / 16 Passed

Repository: crdant/compound-engineering-plugin
Path: skills/ce-optimize/SKILL.md
Commit: f794948

Reviewed: 3 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.