skill-creator

name	skill-creator
description	Create new skills, modify and improve existing skills, and measure skill performance. Use when users want to create a skill from scratch, edit, or optimize an existing skill, run evals to test a skill, benchmark skill performance with variance analysis, or optimize a skill's description for better triggering accuracy.
license	Apache-2.0

Skill Creator

A skill for creating new skills and iteratively improving them.

At a high level, the process of creating a skill goes like this:

• Decide what you want the skill to do and roughly how it should do it
• Write a draft of the skill
• Create a few test prompts and run claude-with-access-to-the-skill on them
• Help the user evaluate the results both qualitatively and quantitatively
  • While the runs happen in the background, draft some quantitative evals if there aren't any (if there are some, you can either use as is or modify if you feel something needs to change about them). Then explain them to the user (or if they already existed, explain the ones that already exist)
  • Use the eval-viewer/generate_review.py script to show the user the results for them to look at, and also let them look at the quantitative metrics
• Rewrite the skill based on feedback from the user's evaluation of the results (and also if there are any glaring flaws that become apparent from the quantitative benchmarks)
• Repeat until you're satisfied
• Expand the test set and try again at larger scale

Your job when using this skill is to figure out where the user is in this process and then jump in and help them progress through these stages. So for instance, maybe they're like "I want to make a skill for X". You can help narrow down what they mean, write a draft, write the test cases, figure out how they want to evaluate, run all the prompts, and repeat.

On the other hand, maybe they already have a draft of the skill. In this case you can go straight to the eval/iterate part of the loop.

Of course, you should always be flexible and if the user is like "I don't need to run a bunch of evaluations, just vibe with me", you can do that instead.

Then after the skill is done (but again, the order is flexible), you can also run the skill description improver, which we have a whole separate script for, to optimize the triggering of the skill.

Reading and writing skill files — host.skills.*

You author skills via the host.skills SDK in the repl tool (not the python tool — it's a separate stdlib-only control-plane kernel that shares your workspace cwd; full API reference in the customize skill). The four calls you need:

host.skills.list()                          # all skills + drafts
host.skills.read(name, path="SKILL.md")     # → {"name","path","content"}
host.skills.edit(name, path, content,       # create/overwrite (old_string=None)
                 old_string=None)           #   or str_replace one match
host.skills.publish(name, overwrite=False)  # promote draft → live skill set

Everything below that says "write SKILL.md" / "save test cases to evals/evals.json" / "ship kernel.py" means host.skills.edit(name, <path>, <content>). Iterate with edit(..., old_string=...) for targeted patches instead of re-emitting whole files; confirm with read(...). When the skill is ready, publish(name) makes it loadable by agents — then attach it to a profile with host.agents.attach_skill(profile, name).

Communicating with the user

The skill creator is used by people across a wide range of familiarity with coding jargon — from complete non-programmers to expert developers. Calibrate your language accordingly.

So please pay attention to context cues to understand how to phrase your communication! In the default case, just to give you some idea:

• "evaluation" and "benchmark" are borderline, but OK
• for "JSON" and "assertion" you want to see serious cues from the user that they know what those things are before using them without explaining them

It's OK to briefly explain terms if you're in doubt, and feel free to clarify terms with a short definition if you're unsure if the user will get it.

---

Creating a skill

Capture Intent

Start by understanding the user's intent. The current conversation might already contain a workflow the user wants to capture (e.g., they say "turn this into a skill"). If so, extract answers from the conversation history first — the tools used, the sequence of steps, corrections the user made, input/output formats observed. The user may need to fill the gaps, and should confirm before proceeding to the next step.

1. What should this skill enable Claude to do?
2. When should this skill trigger? (what user phrases/contexts)
3. What's the expected output format?
4. Should we set up test cases to verify the skill works? Skills with objectively verifiable outputs (file transforms, data extraction, code generation, fixed workflow steps) benefit from test cases. Skills with subjective outputs (writing style, art) often don't need them. Suggest the appropriate default based on the skill type, but let the user decide.

Interview and Research

Proactively ask questions about edge cases, input/output formats, example files, success criteria, and dependencies. Wait to write test prompts until you've got this part ironed out.

Check available MCPs - if useful for research (searching docs, finding similar skills, looking up best practices), research in parallel via subagents if available, otherwise inline. Come prepared with context to reduce burden on the user.

Write the SKILL.md

Based on the user interview, fill in these components:

• name: Skill identifier
• description: When to trigger, what it does. This is the primary triggering mechanism - include both what the skill does AND specific contexts for when to use it. All "when to use" info goes here, not in the body. Note: Claude tends to "undertrigger" skills -- to not use them when they'd be useful. To combat this, please make the skill descriptions a little bit "pushy". So for instance, instead of "How to build a simple fast dashboard to display internal company data.", you might write "How to build a simple fast dashboard to display internal company data. Make sure to use this skill whenever the user mentions dashboards, data visualization, internal metrics, or wants to display any kind of company data, even if they don't explicitly ask for a 'dashboard.'"
• compatibility: Required tools, dependencies (optional, rarely needed)
• the body: the instructions themselves

Skill Writing Guide

Anatomy of a Skill

skill-name/
├── SKILL.md (required)
│   ├── YAML frontmatter (name, description required)
│   └── Markdown instructions
├── kernel.py / kernel.R (optional) - Auto-injected into the live repl on load
└── Bundled Resources (optional)
    ├── scripts/    - Executable code for deterministic/repetitive tasks
    ├── references/ - Docs loaded into context as needed
    └── assets/     - Files used in output (templates, icons, fonts)

Kernel plugins (kernel.py / kernel.R)

If the skill's workflow depends on reusable helper functions, ship them as kernel.py (and/or kernel.R) at the skill root. When any agent calls skill({skill: <name>}), that file is executed in its persistent python/R kernel and the tool result reports which top-level names were defined — so SKILL.md can say "call annotate_df(df)" and the function already exists.

Guidance:

• Top-level definitions only: functions, imports, and literal constants — top-level class statements and decorators are rejected by the validator (define classes inside a factory function if needed). Code runs on every skill load and on kernel restart — keep it idempotent and fast; no network calls or heavy imports at module scope.
• Default argument values must be literals — def f(url=MY_CONSTANT) gets the whole file rejected. Wrap constants with an explicit is None check: def f(url=None): then if url is None: url = MY_CONSTANT (not url = url or MY_CONSTANT, which also replaces 0, "", []).
• Do not use _-prefixed top-level names — they are reserved by the sidecar loader and will cause the entire kernel.py to be rejected.
• Defer third-party imports to inside function bodies. The skeleton python env ships stdlib + a small starter set (numpy, pandas, scipy, matplotlib, seaborn, pillow), so e.g. import requests at module scope surfaces a load error on every fresh kernel. Import errors don't fail the skill load — the agent sees the traceback and can manage_packages then re-load.
• Keep it small and self-contained. kernel.py cannot import from the skill's scripts/ dir (it's exec'd into a per-skill kernel namespace with the skill dir not on sys.path — that import is the classic load error). If a helper wants more than ~100 lines, trim it to the core operations the agent actually calls; scripts/ is for standalone CLI tools run via bash, not for backing the sidecar.
• Namespacing (how your helpers coexist with other skills'): each kernel.py/kernel.R runs in its own namespace, and its public names are mirrored to the shared session — bare when no earlier-loaded skill owns the name (first-owner-wins), and normally also under a stable qualified alias (typically <skill-slug>__<name>; if even the alias name is squatted the load report says which form survives). If another skill already owns a bare name, the load report tells the agent the exact alias to call — you don't need globally unique helper names, but distinctive ones avoid the aliased path. Two semantics to know: your functions always call your same-file helpers and constants (another skill or the agent rebinding the bare name won't redirect them), and globals() inside your functions is the skill's namespace — reads AND 'x' in globals() membership checks both see agent/session names through the fallback (so membership matches what a read would find, and serializers like cloudpickle capture what your functions actually use); keys()/iteration list only your own definitions.

Minimal example:

# kernel.py
import pandas as pd  # starter-set package — OK at module scope

def annotate_df(df: pd.DataFrame, gene_col: str = "gene") -> pd.DataFrame:
    """Attach HGNC symbols; see SKILL.md ## Workflow step 3."""
    import requests  # not in starter set — defer to function body
    ...
    return df

When the gate probe can judge, the host.skills.edit(name, "kernel.py", src) result carries sidecar_gate: {ok, error?} — the same structural gate the load path runs — so a reject surfaces immediately. When the probe can't judge (interpreter unavailable, or the source doesn't parse under the host's interpreter — possible version drift), the key is absent and note says why. Iterate against that; don't load the skill just to test the sidecar. host.skills.publish refuses only on a structural reject.

Progressive Disclosure

Skills use a three-level loading system:

1. Metadata (name + description) - Always in context (~100 words)
2. SKILL.md body - In context whenever skill triggers (<500 lines ideal)
3. Bundled resources - As needed (unlimited, scripts can execute without loading)

These word counts are approximate and you can feel free to go longer if needed.

Key patterns:

• Keep SKILL.md under 500 lines; if you're approaching this limit, add an additional layer of hierarchy along with clear pointers about where the model using the skill should go next to follow up.
• Reference files clearly from SKILL.md with guidance on when to read them
• For large reference files (>300 lines), include a table of contents

Domain organization: When a skill supports multiple domains/frameworks, organize by variant:

cloud-deploy/
├── SKILL.md (workflow + selection)
└── references/
    ├── aws.md
    ├── gcp.md
    └── azure.md

Claude reads only the relevant reference file.

Principle of Lack of Surprise

This goes without saying, but skills must not contain malware, exploit code, or any content that could compromise system security. A skill's contents should not surprise the user in their intent if described. Don't go along with requests to create misleading skills or skills designed to facilitate unauthorized access, data exfiltration, or other malicious activities. Things like a "roleplay as an XYZ" are OK though.

Writing Patterns

Prefer using the imperative form in instructions.

Defining output formats - You can do it like this:

## Report structure

ALWAYS use this exact template:

# [Title]

## Executive summary

## Key findings

## Recommendations

Examples pattern - It's useful to include examples. You can format them like this (but if "Input" and "Output" are in the examples you might want to deviate a little):

## Commit message format

**Example 1:**
Input: Added user authentication with JWT tokens
Output: feat(auth): implement JWT-based authentication

Writing Style

Try to explain to the model why things are important in lieu of heavy-handed musty MUSTs. Use theory of mind and try to make the skill general and not super-narrow to specific examples. Start by writing a draft and then look at it with fresh eyes and improve it.

Test Cases

After writing the skill draft, come up with 2-3 realistic test prompts — the kind of thing a real user would actually say. Share them with the user: [you don't have to use this exact language] "Here are a few test cases I'd like to try. Do these look right, or do you want to add more?" Then run them.

Save test cases to evals/evals.json. Don't write expectations yet — just the prompts. You'll draft expectations in the next step while the runs are in progress.

{
  "skill_name": "example-skill",
  "evals": [
    {
      "id": 1,
      "prompt": "User's task prompt",
      "expected_output": "Description of expected result",
      "files": []
    }
  ]
}

See references/schemas.md for the full schema (including the expectations field, which you'll add later).

Running and evaluating test cases

This section is one continuous sequence — don't stop partway through. Do NOT use /skill-test or any other testing skill.

Put results in <skill-name>-workspace/ as a sibling to the skill directory. Within the workspace, organize results by iteration (iteration-1/, iteration-2/, etc.) and within that, each test case gets a directory (eval-0/, eval-1/, etc.). Don't create all of this upfront — just create directories as you go.

Step 1: Spawn all runs (with-skill AND baseline) in the same turn

For each test case, spawn two subagents in the same turn — one with the skill, one without. This is important: don't spawn the with-skill runs first and then come back for baselines later. Launch everything at once so it all finishes around the same time.

With-skill run:

Execute this task:
- Skill path: <path-to-skill>
- Task: <eval prompt>
- Input files: <eval files if any, or "none">
- Save outputs to: <workspace>/iteration-<N>/eval-<ID>/with_skill/outputs/
- Outputs to save: <what the user cares about — e.g., "the .docx file", "the final CSV">

Baseline run (same prompt, but the baseline depends on context):

• Creating a new skill: no skill at all. Same prompt, no skill path, save to without_skill/outputs/.
• Improving an existing skill: the old version. Before editing, snapshot the skill (cp -r <skill-path> <workspace>/skill-snapshot/), then point the baseline subagent at the snapshot. Save to old_skill/outputs/.

Write an eval_metadata.json for each test case (assertions can be empty for now). Give each eval a descriptive name based on what it's testing — not just "eval-0". Use this name for the directory too. If this iteration uses new or modified eval prompts, create these files for each new eval directory — don't assume they carry over from previous iterations.

{
  "eval_id": 0,
  "eval_name": "descriptive-name-here",
  "prompt": "The user's task prompt",
  "assertions": []
}

Step 2: While runs are in progress, draft expectations

Don't just wait for the runs to finish — you can use this time productively. Draft quantitative expectations for each test case and explain them to the user. If expectations already exist in evals/evals.json, review them and explain what they check.

Good expectations are objectively verifiable and have descriptive names — they should read clearly in the benchmark viewer so someone glancing at the results immediately understands what each one checks. Subjective skills (writing style, design quality) are better evaluated qualitatively — don't force expectations onto things that need human judgment.

Update evals/evals.json with the expectations once drafted, and the eval_metadata.json files with the assertions. Also explain to the user what they'll see in the viewer — both the qualitative outputs and the quantitative benchmark.

Step 3: As runs complete, capture timing data

When each subagent task completes, you receive a notification containing total_tokens and duration_ms. Save this data immediately to timing.json in the run directory:

{
  "total_tokens": 84852,
  "duration_ms": 23332,
  "total_duration_seconds": 23.3
}

This is the only opportunity to capture this data — it comes through the task notification and isn't persisted elsewhere. Process each notification as it arrives rather than trying to batch them.

Step 4: Grade, aggregate, and launch the viewer

Once all runs are done:

1. Grade each run — spawn a grader subagent (or grade inline) that reads agents/grader.md and evaluates each expectation against the outputs. Save results to grading.json in each run directory. The grading.json expectations array must use the fields text, passed, and evidence (not name/met/details or other variants) — the viewer depends on these exact field names. For expectations that can be checked programmatically, write and run a script rather than eyeballing it — scripts are faster, more reliable, and can be reused across iterations.

2. Aggregate into benchmark — run the aggregation script from the skill-creator directory:

   python -m scripts.aggregate_benchmark <workspace>/iteration-N --skill-name <name>
   

   This produces benchmark.json and benchmark.md with pass_rate, time, and tokens for each configuration, with mean ± stddev and the delta. If generating benchmark.json manually, see references/schemas.md for the exact schema the viewer expects. Put each with_skill version before its baseline counterpart.

3. Do an analyst pass — read the benchmark data and surface patterns the aggregate stats might hide. See agents/analyzer.md (the "Analyzing Benchmark Results" section) for what to look for — things like expectations that always pass regardless of skill (non-discriminating), high-variance evals (possibly flaky), and time/token tradeoffs.

4. Launch the viewer with both qualitative outputs and quantitative data:

   nohup python <skill-creator-path>/eval-viewer/generate_review.py \
     <workspace>/iteration-N \
     --skill-name "my-skill" \
     --benchmark <workspace>/iteration-N/benchmark.json \
     > /dev/null 2>&1 &
   VIEWER_PID=$!
   

   For iteration 2+, also pass --previous-workspace <workspace>/iteration-<N-1>.

   Cowork / headless environments: If webbrowser.open() is not available or the environment has no display, use --static <output_path> to write a standalone HTML file instead of starting a server. Feedback will be downloaded as a feedback.json file when the user clicks "Submit All Reviews". After download, copy feedback.json into the workspace directory for the next iteration to pick up.

Note: please use generate_review.py to create the viewer; there's no need to write custom HTML.

5. Tell the user something like: "I've opened the results in your browser. There are two tabs — 'Outputs' lets you click through each test case and leave feedback, 'Benchmark' shows the quantitative comparison. When you're done, come back here and let me know."

What the user sees in the viewer

The "Outputs" tab shows one test case at a time:

• Prompt: the task that was given
• Output: the files the skill produced, rendered inline where possible
• Previous Output (iteration 2+): collapsed section showing last iteration's output
• Formal Grades (if grading was run): collapsed section showing expectation pass/fail
• Feedback: a textbox that auto-saves as they type
• Previous Feedback (iteration 2+): their comments from last time, shown below the textbox

The "Benchmark" tab shows the stats summary: pass rates, timing, and token usage for each configuration, with per-eval breakdowns and analyst observations.

Navigation is via prev/next buttons or arrow keys. When done, they click "Submit All Reviews" which saves all feedback to feedback.json.

Step 5: Read the feedback

When the user tells you they're done, read feedback.json:

{
  "reviews": [
    {