Skip to content

sickn33/agentic-awesome-skills / plugins/aasb-skill-author/skills/skill-creator

skill-creator

Verified

Create and review reusable skill drafts with bundled Python scaffolding, basic frontmatter validation and ZIP packaging helpers. Use for skill authoring and packaging; verify target-host requirements separately.

FileKindSize
LICENSE.txtmarkdown11.1 KB
README.mdmarkdown2.2 KB
SKILL.mdskill md1.6 KB
references/detailed-guide.mdmarkdown3.7 KB
references/output-patterns.mdmarkdown1.8 KB
references/workflows.mdmarkdown818 B
scripts/init_skill.pyscript11.7 KB
scripts/package_skill.pyscript5.1 KB
scripts/quick_validate.pyscript4.4 KB

references/detailed-guide.md

View on GitHub
# Skill Creator Workflow

Modified for the AAS catalog: corrected bundled workflow, validation and packaging safeguards (2026-10-23).

## 1. Establish scope

Use the user's request and repository instructions to determine the skill name, purpose, activation examples, target host and output directory. Ask only for missing decisions that affect the result. Skill creation, global installation and publication are separate actions; honor the requested boundary. Do not infer host support from the existence of a configuration directory.

## 2. Create a draft

Read the repository's current contribution template first. If a generic scaffold is useful, invoke the bundled `scripts/init_skill.py` with Python 3.10+ from the intended workspace:

```bash
python3 <skill-dir>/scripts/init_skill.py my-skill --path skills
```

The skill ID must be lowercase hyphenated words, at most 64 characters. The output directory must be within the current workspace and must not already exist. The helper creates SKILL.md, one placeholder Python script, a reference and an asset placeholder. This is a draft, not a finished or validated skill.

Write YAML with a YAML serializer or carefully quoted static values; do not interpolate arbitrary user descriptions into shell `sed` programs. Add examples, prerequisites, dependencies, risk, provenance and limitations appropriate to the repository. Retain licenses and explain local changes to imported material. Remove template TODOs and unneeded placeholder files. Do not invent unavailable templates, scripts or successful outputs.

## 3. Review every bundled file

Keep concise activation/workflow instructions in SKILL.md and supporting details in references. Read every script before deciding whether to execute it. Inspect input validation, path handling, credentials, remote destinations, side effects, output overwrites, failure behavior and resource bounds. Distinguish deterministic checks from heuristic judgments. Do not run code merely to satisfy a template.

See [output-patterns.md](output-patterns.md) and [workflows.md](workflows.md) for authoring patterns. Other skills can be considered only when installed, relevant and authorized; no automatic dependency or delegation is implied.

## 4. Validate and exercise representative behavior

```bash
python3 <skill-dir>/scripts/quick_validate.py skills/my-skill
```

The generic quick validator requires PyYAML. Its allowed fields are deliberately small and may differ from a host or repository schema. Run the current repository's required validators separately. A frontmatter pass says nothing about semantic quality or runtime safety. Test representative user requests in the actual target host when available; state any unavailable checks honestly. Do not invent fixed word-count standards or portray style preferences as universal host requirements.

## 5. Package if requested

```bash
python3 <skill-dir>/scripts/package_skill.py skills/my-skill dist
```

Choose a new output archive outside the skill directory. Packaging does not run bundled scripts. The archive is a ZIP using the `.skill` suffix; target-host installation support must be verified separately. Inspect archive contents and remove secrets before sharing. Exclusions are convenience safeguards, not a complete secret scan.

## 6. Install or hand off within scope

For local/global installation, consult the selected host's current documented discovery paths rather than assuming `.codex/skills`, `.copilot/skills` or another path works everywhere. Inspect the destination and preserve existing work; do not force-overwrite a directory or symlink. Verify discovery with the actual host. If only a draft was requested, deliver its location and validation limits without installing, committing, publishing or contacting anyone.
reverse-engineer-anythingmorluto/reaReverse engineer native, managed, Electron/JavaScript, packaged, firmware, and browser targets with REA, and build agent environments (AgentEnv bundles) of web applications from a live instance or a capture, and of Linux desktop apps with a task from a request. Use shipped-artifact or requested runtime evidence to explain features, compare versions, decompile code, or guide a reconstruction. Skip REA for ordinary source-repository architecture analysis.Stars: 472kUpdated Claude CodeCodexVerifiedInstall
diagnosing-superpowersobra/superpowersUse when a superpowers session went wrong and your human partner wants to know why — repeated work, ignored plans, stumbles, poor results, a skill that didn't fire, "it took too long", "why is it so expensive", "what is it doing" — or wants to build a bug report for the superpowers maintainers, for the current session or a past one identified by id or path, on any harness.Stars: 322kUpdated Claude CodeCodexVerifiedInstall
dispatching-parallel-agentsobra/superpowersUse when facing 2+ independent tasks that can be worked on without shared state or sequential dependenciesStars: 322kUpdated Claude CodeCodexVerifiedInstall
executing-plansobra/superpowersUse when executing an implementation plan in the current session as the implementer yourself — your human partner chose inline execution, or no subagent tool is availableStars: 322kUpdated Claude CodeCodexVerifiedInstall