| name | ingest-tool |
| description | Read source code of any tool and generate an AuthorAgent skill from it |
| author | AuthorAgent |
| version | 1.0.0 |
| triggers | ["ingest tool","create skill from","import tool","make a skill","convert to skill","read this code","analyze this tool"] |
| permissions | ["file:read","file:write"] |
Tool Ingestion โ Code to Skill Converter
Read source code of any external tool and generate a proper AuthorAgent SKILL.md file.
Process
- Read the source code provided by the user (file path or pasted code)
- Analyze the tool:
- Purpose and domain (writing, formatting, marketing, analysis)
- Inputs: arguments, config files, stdin
- Outputs: files, stdout, formats
- Language: Python, JavaScript, HTML, etc.
- CLI interface and flags (if any)
- Dependencies required
- Generate SKILL.md with:
- YAML frontmatter: name, description, triggers, permissions
- How the tool works
- Input/output documentation
- Example commands or workflows
- How AuthorAgent should invoke or reference it
- Save to
skills/author/[tool-name]/SKILL.md (or marketing/core as appropriate)
SKILL.md Format
---
name: tool-name-kebab-case
description: One-line description of what the tool does
author: AuthorAgent (generated from [original tool])
version: 1.0.0
triggers:
- "keyword 1"
- "keyword 2"
permissions:
- file:read
- file:write
---
[How it works, when to use it, example commands]
Rules
- NEVER invent capabilities the tool doesn't have
- Note required dependencies (Python version, npm packages)
- If the tool is GUI-only, note it requires manual interaction
- If the tool has a CLI, document exact command syntax
- Keep triggers practical โ words users would actually type
- Name must be kebab-case, unique, and descriptive