{"manifest":{"name":"Skill Classifier","version":"1.0.0","description":"Classify skills by type, requirements, and risk level. First step in the 3-pass verification pipeline.","tags":["verification","classifier","meta","toolkit"],"standard":"agentskills.io","standard_version":"1.0","content_checksum":"557f0eb8c9c487b283c33f07ff945f80b23a5227db39219d709e10e988c3089c","bundle_checksum":null,"metadata":{},"files":[]},"files":{"SKILL.md":"# Skill Classifier — Verification Toolkit\n\n> **Purpose:** Teach an AI agent how to classify a SkillSlap skill by type, requirements, and risk level.\n> This is the first step in the 3-pass verification pipeline.\n\n---\n\n## 1. Overview\n\nThe Skill Classifier analyzes a skill's markdown content and produces a structured classification that determines:\n\n1. **What kind of skill it is** (documentation, API workflow, code generation, etc.)\n2. **What it requires** to be properly verified (API access, sandbox, browser, specific tools)\n3. **How risky it is** to execute (safe, moderate, high, critical)\n\nThis classification feeds into the verification pipeline to determine which checks to run.\n\n---\n\n## 2. Input\n\nYou receive a skill with:\n\n```json\n{\n  \"title\": \"string\",\n  \"description\": \"string | null\",\n  \"content\": \"string (markdown)\",\n  \"tags\": [\"string\"],\n  \"version\": \"string\"\n}\n```\n\n---\n\n## 3. Classification Schema\n\n### 3a. Skill Type\n\n| Type | Description | Examples |\n|------|-------------|----------|\n| `documentation` | Reference material, guides, how-tos | API docs, tutorials, READMEs |\n| `api_workflow` | Interacts with external APIs | REST clients, webhook handlers |\n| `code_generation` | Generates or transforms code | Scaffolding tools, transpilers |\n| `agent_instructions` | Instructions for an AI agent to follow | Prompt templates, agent workflows |\n| `tool_config` | Configuration for tools, CLIs, environments | Docker configs, CI/CD pipelines |\n| `mixed` | Combines multiple types | Full-stack guides with API + code |\n\n### 3b. Requirements\n\n| Requirement | When to Set True |\n|------------|-----------------|\n| `api_access` | Skill references HTTP endpoints, webhooks, or external services |\n| `code_sandbox` | Skill includes executable code that should be run in isolation |\n| `browser_rendering` | Skill produces HTML/CSS output that needs visual verification |\n| `specific_tools` | List any CLI tools, runtimes, or libraries the skill requires |\n\n### 3c. Risk Level\n\n| Level | Criteria |\n|-------|----------|\n| `safe` | Read-only, no side effects, documentation only |\n| `moderate` | Makes API calls but no destructive operations |\n| `high` | Writes data, modifies state, or accesses credentials |\n| `critical` | System-level operations, shell commands, or credential management |\n\n---\n\n## 4. Classification Process\n\n1. **Read the skill content** carefully\n2. **Identify the primary type** — what does the skill mostly do?\n3. **Check for requirements** — does it need APIs? Code execution? A browser?\n4. **Assess risk** — what could go wrong if an agent follows these instructions blindly?\n5. **Provide reasoning** — 1-2 sentences explaining your classification\n\n---\n\n## 5. Output Format\n\n```json\n{\n  \"type\": \"api_workflow\",\n  \"requirements\": {\n    \"api_access\": true,\n    \"code_sandbox\": false,\n    \"browser_rendering\": false,\n    \"specific_tools\": [\"curl\", \"jq\"]\n  },\n  \"risk_level\": \"moderate\",\n  \"reasoning\": \"This skill instructs agents to make HTTP requests to external APIs with authentication headers. No destructive operations but does write data via POST requests.\"\n}\n```\n\n---\n\n## 6. Classification Heuristics\n\n### Indicators of Type\n\n- **documentation**: Headers like \"Overview\", \"Reference\", no executable blocks\n- **api_workflow**: HTTP method blocks (`GET`, `POST`), URL patterns, `Authorization` headers\n- **code_generation**: Code blocks with language tags, file output patterns, scaffolding steps\n- **agent_instructions**: Imperative mood (\"Do X\", \"Then Y\"), decision trees, tool usage patterns\n- **tool_config**: Config file formats (YAML, TOML, JSON), environment variables, setup steps\n- **mixed**: Multiple indicators from different categories\n\n### Indicators of Risk\n\n- **safe**: \"This guide explains...\", no code blocks, no URLs\n- **moderate**: API calls with read operations, query parameters\n- **high**: `POST`/`PUT`/`DELETE` with bodies, file writes, database operations\n- **critical**: `rm`, `sudo`, shell commands, credential storage, `eval`, system calls\n\n---\n\n## 7. Edge Cases\n\n- **Skills with code examples** that are illustrative (not meant to be executed) → `documentation`, `safe`\n- **Skills that teach security testing** → classify based on the operations described, not the topic\n- **Skills referencing external tools without including them** → still note in `specific_tools`\n- **Empty or minimal skills** → `documentation`, `safe`, note in reasoning"}}