skill-improve

skill-improve is a skill for Claude Code from ouzlifaneyassine1-dot/onyx-engine. It costs 68 tokens per session (787 once invoked), scanned A, a copy of skill-improve, MIT.

A workflow for improving Onyx skills by comparing their behavior before and after proposed changes.

In plain words
What is it for?
It runs skills against behavioral test cases and keeps the version that performs better.
Why use it?
It helps identify whether an edit actually fixes failing behavior instead of relying only on reading the instructions.

Skill for Claude Code

Written for Claude Code: allowed-tools in frontmatter. Also seen: mentions subagents; mentions Claude Code; mentions Codex.

Part of the onyx plugin — 77 skills, 1 command, 2 hooks, 1 MCP server shipped together

Good fit It runs skills against behavioral test cases and keeps the version that performs better.

Compare 6 skills from other repositories ↓
Install with agentmods
npx agentmods add skills/ouzlifaneyassine1-dot/onyx-engine/skill-improve
Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

Any agent
npx skills add ouzlifaneyassine1-dot/onyx-engine --skill skill-improve
Clone the repo
git clone --depth 1 https://github.com/ouzlifaneyassine1-dot/onyx-engine

Made for: Claude Code.

Or install onyx, the plugin that ships this one along with the rest of its 77 skills, 1 command, 2 hooks, 1 MCP server.

Wrote this? Show the measurements

A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.

agentmods badge for skill-improve

README.md
[![agentmods](https://agentmods.dev/badge/skills/ouzlifaneyassine1-dot/onyx-engine/skill-improve/github.svg)](https://agentmods.dev/skills/ouzlifaneyassine1-dot/onyx-engine/skill-improve)
Your own site
<a href="https://agentmods.dev/skills/ouzlifaneyassine1-dot/onyx-engine/skill-improve"><img src="https://agentmods.dev/badge/skills/ouzlifaneyassine1-dot/onyx-engine/skill-improve/github.svg" alt="Measured on agentmods" height="20"></a>

Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.

agentmods 80×15 button for skill-improve

Your own site · 80×15
<a href="https://agentmods.dev/skills/ouzlifaneyassine1-dot/onyx-engine/skill-improve"><img src="https://agentmods.dev/badge/skills/ouzlifaneyassine1-dot/onyx-engine/skill-improve.svg" alt="Reviewed on agentmods" width="80" height="20"></a>
Per session 68 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 787 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. A grade says what 26 rules found in the file — not that it is safe.
Origin 97% copy Near-identical to another mod in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5.1 $0.00068 $0.00787
Opus 5 $0.00034 $0.00394
Sonnet 5 $0.00014 $0.00157
Haiku 4.5 $0.00007 $0.00079

Measured 7d ago against content hash c9ca87ebd4ca, method: parsed. Prices are Anthropic first-party input rates as of 2026-09-10, from the pricing page.

Security

Grade A, and why

skill-improve scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 7d ago.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

Origin

This is a copy

97% identical to skill-improve — 2 lines differ, which has more behind it and is treated as the original. This page carries a canonical link to it rather than competing with it.

skills/workflow/skill-improve/SKILL.md · 89 lines

How it starts

The opening of the file, as written. The whole thing — 89 lines — stays where its author put it; the contents beside it link to each section on GitHub.

/skill-improve — Iterate on a Skill With Eval Harness

Inspired by anthropics/skills skill-creator/. Use when a skill's behavioral spec is failing assertions or producing low-quality output.

When to use this vs. /skill-test

  • /skill-test spec reasons over text. Cheap, fast, lossy.
  • /skill-improve actually runs the skill in parallel subagents with-vs-without proposed changes. Expensive, accurate.

Run this when /skill-test spec flags issues you can't fix by reading the skill alone.

Steps

1. Pick the skill + the spec

Ask the user:

  • Skill name. Resolves to skills/<category>/<name>/SKILL.md.
  • Test cases to focus on. Default: all ## Case blocks in tests/specs/<name>.md.

2. Establish a baseline

For each Case:

  1. Spawn a subagent (Task tool, general-purpose) with the current skill body in context.
  2. Give it the Case's Input + Fixture.
  3. Capture the tool calls it makes and the diff it produces.
  4. Score against the Case's Assertions.

Save outputs to tests/runs/<skill-name>/baseline/case-<N>/.

3. Propose changes

Read the failing Cases. Identify the gap between what the skill says and what the agent did. Common gaps:

  • Skill names a tool but the agent picked a different one (clarify the trigger).
  • Skill assumes a fixture detail the agent missed (add explicit step to confirm).
  • Skill's "May I" wording is too generic for the agent to infer the right ask.

Draft a revised SKILL.md. May I write it to tests/runs/<skill-name>/proposed/SKILL.md?

4. Run the proposed version

Repeat step 2 with the proposed SKILL.md. Save to tests/runs/<skill-name>/proposed/case-<N>/.

5. Compare and decide

For each Case, score:

  • Assertions passed (proposed vs. baseline).
  • Tool-call efficiency (fewer tools = better, all else equal).
  • Hallucination / unwanted ops (penalize).

Output:

Case 1 (Happy):     baseline 4/6  proposed 6/6  ✓ ship proposed
Case 2 (Failure):   baseline 3/4  proposed 4/4  ✓ ship proposed
Case 3 (Edge):      baseline 3/3  proposed 3/3  = no change

Read the full file on GitHub · 89 lines

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. 7d ago First seen · 89 lines · 68 tokens per session scan A c9ca87ebd4ca

Subscribe to this mod's changes

skill-improve is a skill published in the GitHub repository ouzlifaneyassine1-dot/onyx-engine (0 stars, last pushed 2mo ago), licensed MIT. It adds 68 tokens to every session and 787 once invoked, about $0.0003 per session on Opus 5. A static security scan graded it A with 0 findings. It is 97% identical to skill-improve, differing in 2 lines, and is treated as a copy.

Related

Other skills, from other repositories

trulens-evaluation-workflow

Systematically evaluate your LLM application with TruLens.

truera/trulens · 18 tokens

agent-builder

Load before calling build-agent for a new or existing n8n Agent. Governs prerequisite creation, faithful handoff of the user's request, agent targeting across turns, builder questions, testing, and publishing. Use directly for routine follow-ups when the conversation already targets an Agent; rerun intent-recognition…

n8n-io/n8n · 73 tokens

playwright-cli

A command-line tool for controlling Chromium, Firefox, and WebKit browsers, including navigation, page interaction, screenshots, PDFs, and recorded actions.

UnicomAI/wanwu · 76 tokens

config-evals

Builds and maintains configuration-based evaluations on a workflow with the eval-config tool. Use when the user asks to set up, add, view, change, or remove an evaluation, score, grade, or judge a workflow's output, or measure answer quality against a test dataset. This is the only eval form Instance AI handles — it…

n8n-io/n8n · 80 tokens

ccg-workflow

How to run a non-trivial change end to end with the CCG role tools (ccganalyze / ccgdesign / ccgbuild / ccgdebug / ccgoptimize / ccgreview / ccgtest) and the verify- quality gates. Use when a task is a whole feature, a refactor, a bug whose cause is not yet known, or any change large enough that one straight-through…

fengshao1227/ccg-workflow · 97 tokens

flow-next-qa

Live-app QA pass derived from the spec. Drives the running app, files P0/P1/P2 findings with evidence, emits a YES or NO qaverdict receipt.

gmickel/flow-next · 40 tokens