skills.rest
tcytan
tcytanCOMMUNITY
Explore all skills by this author →
WebsiteGitHub

skill-iter-tune

Execute, evaluate, and improve workflow skills until a score threshold is reached.

1|Updated May 12, 2026
Search Tags:#prompt engineering#workflow orchestration#quality evaluation#iterative refinement#skill tuning#agentic improvement#ccw cli
One-click install
npx skills add https://github.com/tcytan/tecton --skill skill-iter-tune
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: skill-iter-tune
Source: https://github.com/tcytan/tecton/tree/main/.claude/skills/skill-iter-tune
Command: npx skills add https://github.com/tcytan/tecton --skill skill-iter-tune

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

It solves the challenge of improving an existing skill by repeatedly running it, objectively evaluating its outputs, and applying targeted changes until it meets a quality threshold.

Core Features & Use Cases

  • Execute–Evaluate–Improve Loop: Runs the target skill(s), scores the results across multiple quality dimensions, and applies improvements based on the evaluator’s suggestions.
  • Single or Chain Tuning: Tunes one skill in isolation or tunes a sequence of skills where each skill’s output feeds into the next.
  • Versioned Iterations with Reports: Snapshots each iteration, stores evaluation and change logs, and produces a final report with score progression.

Quick Start

Use the skill iter-tune to tune your workflow skill(s by running it on a local skill directory with a defined test scenario, for example: ccw skill "skill-iter-tune" ".claude/skills/my-skill" --scenario "validate that the skill produces correct artifacts for a typical request".

Frequently Asked Questions about skill-iter-tune

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I iteratively improve a workflow skill until it reaches a quality threshold?▼

Iterative skill improvement is achieved by executing the target skill, evaluating its output quality across multiple dimensions, and applying agent-driven code changes until a composite score threshold or iteration limit is reached.

Can I tune multiple skills that run sequentially and share context?▼

Chain-mode tuning supports sequentially running multiple skills that share context through produced artifacts, allowing you to evaluate and improve the entire workflow chain together.

What do I need to set up before tuning a skill locally?▼

Workspace setup and skill path validation via a SKILL.md file are required before tuning, ensuring deterministic execution orchestration through the ccw CLI.

How does the evaluation mechanism work during prompt engineering refinement?▼

The evaluation mechanism uses Claude for execution and Gemini for evaluation, parsing structured JSON results to apply targeted code improvements based on the evaluator's suggestions.

How are iteration results tracked during workflow orchestration tuning?▼

Each iteration snapshots evaluation logs and change logs as per-iteration artifacts, ultimately producing a final report that documents the score progression throughout the tuning process.

Related Skills

Explore More Software Engineering Skills→
PeiiiiPeiiii
111

alpha

Provide a minimal local test skill for validating installation and discovery workflows.

Community
Basic
tae0ytae0y
14

test-design-reviewer

Scores test quality using Dave Farley's eight properties and outputs a structured report.

Community
Advanced
teklinertekliner
10

multi-agent-orchestrator

Coordinate parallel CLI agents to evaluate competing solutions and declare a winner.

Official
Advanced
catlog22catlog22
76

skill-iter-tune

Run execute-evaluate-improve loops on skill definitions and generated artifacts.

Community
Advanced