Acong Layout Demo

Kitchen-sink SKILL.md demonstrating the full recommended skill directory layout — multi-file progressive disclosure (FORMS.md + REFERENCE.md), a `references/...

Security Scan

This is a structural demo skill whose files, instructions, and small helper scripts are coherent with its stated purpose and do not request extra credentials or perform unexpected network or privileged actions.

Purpose & Capability

The name/description (layout demo) matches the files and behavior: documentation, example assets, and small deterministic helper scripts. There are no unrelated env vars, binaries, or config paths requested.

Instruction Scope

SKILL.md explains progressive-disclosure usage and only directs the agent to read documentation files or run the included scripts on demand. The scripts operate on explicit arguments and do not instruct the agent to read arbitrary system state, secrets, or send data to external endpoints.

Install Mechanism

No install spec (instruction-only) is present. The skill includes two small local scripts; nothing is downloaded or written to disk by an installer.

Credentials

No environment variables, credentials, or config paths are required. The included assets/config.example.yaml is only an example file; nothing in the repo asks for secrets or unrelated service keys.

Persistence & Privilege

always is false and the skill does not request permanent presence or modify other skills/system settings. Autonomous invocation is allowed by default but not accompanied by elevated privileges or broad credential access.

Assessment

This repository is a harmless layout/demo. The only executable helpers are scripts/hello.sh (prints a greeting) and scripts/validate.py (checks a file exists and is non-empty; it does not print file contents or reach the network). If you allow an agent to run Level 3 scripts, be mindful that an agent could be instructed to invoke those scripts with arbitrary filesystem paths — validate.py only reports existence/size, but avoid passing sensitive file paths to any automated agent unless you trust it and the runtime environment.

A reference skill that exercises every structural convention published by Anthropic + the openclaw / vercel-labs community. Use it as a template when bootstrapping non-trivial skills.

When to use

• Authoring a new complex skill and looking for a layout blueprint
• Validating a skill loader / registry / CLI handles nested folders
• Teaching Progressive Disclosure (Level 1/2/3) by example

Progressive Disclosure levels demonstrated

Level	File	Purpose
1 — metadata (always loaded)	SKILL.md frontmatter	name + description
2 — instructions (on trigger)	SKILL.md body (this text)	Overview + when-to-use
3 — references (on demand)	FORMS.md, REFERENCE.md (flat) + references/*.md (nested)	Deep docs, loaded only when needed
3 — scripts (on demand)	scripts/*.sh, scripts/*.py	Deterministic helpers executed via bash
3 — assets (on demand)	assets/*	Templates, schemas, fixtures

Structure map

acong-layout-demo/
├── SKILL.md              ← you are here
├── FORMS.md              ← flat-style reference (Anthropic doc example)
├── REFERENCE.md          ← flat-style API ref
├── references/
│   ├── api-guide.md
│   └── examples.md
├── scripts/
│   ├── hello.sh          ← shell helper
│   └── validate.py       ← python helper
└── assets/
    ├── template.md.tpl   ← text template
    └── config.example.yaml

Usage hints for Claude

1. Need a quick answer? Stay in this file (Level 2).
2. Need to fill a form? Read FORMS.md.
3. Need API spec? Read REFERENCE.md or references/api-guide.md for extended version.
4. Need deterministic validation? Run scripts/validate.py <file>.
5. Need a template? Copy assets/template.md.tpl and substitute placeholders.

Not functional

This skill doesn't do anything — it's a structural demo. Copy its layout when starting a real skill.