skill-iter-tune
Iterative skill tuning via execute-evaluate-improve feedback loop. Uses ccw cli Claude to execute skill, Gemini to evaluate quality, and Agent to apply improvements. Iterates until quality threshold or max iterations. Triggers on "skill iter tune", "iterative skill tuning", "tune skill".
78
73%
Does it follow best practices?
Impact
Pending
No eval scenarios have been run
Critical
Do not install without reviewing
Optimize this skill with Tessl
npx tessl skill review --optimize ./.claude/skills/skill-iter-tune/SKILL.mdQuality
Discovery
100%Based on the skill's description, can an agent find and select it at the right time? Clear, specific descriptions lead to better discovery.
This is a strong skill description that clearly communicates a specific, well-defined capability with explicit trigger terms. It names the concrete tools and workflow steps involved (execute via ccw cli Claude, evaluate via Gemini, improve via Agent) and provides clear 'when to use' guidance. The only minor weakness is that it uses somewhat technical/internal terminology (ccw cli, Gemini, Agent) that may not be meaningful to all users, but this is appropriate for what appears to be a developer-facing skill.
| Dimension | Reasoning | Score |
|---|---|---|
Specificity | Lists multiple specific concrete actions: execute skill via ccw cli Claude, evaluate quality via Gemini, apply improvements via Agent, and iterate until quality threshold or max iterations. The execute-evaluate-improve feedback loop is clearly described. | 3 / 3 |
Completeness | Clearly answers both 'what' (iterative skill tuning via execute-evaluate-improve feedback loop using specific tools) and 'when' (explicit triggers: 'skill iter tune', 'iterative skill tuning', 'tune skill'). | 3 / 3 |
Trigger Term Quality | Includes natural trigger phrases users would say: 'skill iter tune', 'iterative skill tuning', 'tune skill'. These are specific and natural terms a user would use when wanting this functionality. | 3 / 3 |
Distinctiveness Conflict Risk | Highly distinctive with a very specific niche: iterative skill tuning using a particular multi-agent pipeline (ccw cli Claude, Gemini, Agent). The trigger terms are unique and unlikely to conflict with other skills. | 3 / 3 |
Total | 12 / 12 Passed |
Implementation
47%Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.
The skill demonstrates excellent workflow clarity with well-defined phases, termination conditions, error recovery, and checklists. However, it is severely over-engineered for a SKILL.md overview — it duplicates information across multiple representations (architecture diagram, execution flow, data flow) and inlines extensive implementation detail that belongs in the referenced phase files. The result is a verbose document that undermines its own progressive disclosure strategy.
Suggestions
Move the detailed JavaScript pseudocode (preference collection, TodoWrite patterns, termination logic) into the respective phase files, keeping only a brief summary of each in SKILL.md.
Consolidate the three overlapping flow representations (Architecture Overview diagram, Execution Flow section, Data Flow section) into a single concise diagram.
Remove the inline error handling table and coordinator checklist from SKILL.md — these are execution details that belong in phase docs or a separate reference file.
Trim the Interactive Preference Collection section to just describe the config options (Quick/Standard/Thorough + Single/Chain) without the full JavaScript implementation.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The skill is extremely verbose at ~300+ lines. It includes extensive ASCII diagrams, redundant data flow representations (the architecture diagram, execution flow, and data flow sections all convey overlapping information), detailed JavaScript pseudocode for preference collection, TodoWrite patterns, and termination logic that could be dramatically condensed. Much of this is orchestration boilerplate Claude can infer. | 1 / 3 |
Actionability | The skill provides concrete pseudocode for termination logic, preference collection, and iteration loops, plus specific CLI commands like 'ccw cli --tool claude --mode write'. However, the actual execution details are deferred to external phase files (phases/01-05), so the SKILL.md itself is more of a coordination map than executable guidance. The code examples are pseudocode/JavaScript-style rather than fully executable. | 2 / 3 |
Workflow Clarity | The multi-step workflow is clearly sequenced with explicit phases, termination conditions, convergence detection, error handling with retry/rollback strategies, and a coordinator checklist. Validation checkpoints are present (check termination before Phase 4, error budget tracking, snapshot before execute for rollback). | 3 / 3 |
Progressive Disclosure | The skill correctly references external phase documents (phases/01-05.md) with a clear table and one-level-deep references. However, the SKILL.md itself contains far too much inline detail that should live in those phase files — the TodoWrite patterns, interactive preference collection code, termination logic, and error handling tables bloat what should be a concise overview. | 2 / 3 |
Total | 8 / 12 Passed |
Validation
90%Checks the skill against the spec for correct structure and formatting. All validation checks must pass before discovery and implementation can be scored.
Validation — 10 / 11 Passed
Validation for skill structure
| Criteria | Description | Result |
|---|---|---|
allowed_tools_field | 'allowed-tools' contains unusual tool name(s) | Warning |
Total | 10 / 11 Passed | |
- Repository
- catlog22/Claude-Code-Workflow
- Commit
0f8e801
- Reviewed
Table of Contents
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.