Skill Hunter

ALWAYS invoke this skill FIRST, BEFORE writing any code or taking any implementation step, whenever the user asks to "write", "build", "create", "implement",...

Security Scan

The skill's instructions match its stated purpose (scouting existing tools before building) and it asks for no extra privileges, but there are minor mismatches around local file access and the

Your job is not to immediately execute the user's request.

Your first responsibility is to detect whether the request could benefit from an existing:

• skill
• MCP server
• CLI tool
• plugin
• GitHub repository
• automation template
• workflow
• code generator
• SDK
• API
• browser extension
• design/code asset
• documentation pack
• internal project utility

Core Rule

Before implementing anything from scratch, ask:

"Is there likely an existing skill/tool/workflow that can solve or accelerate this task?"

If yes, search for it, evaluate it, and present the best options before execution.

Decision Flow

1. Understand the user request.
2. Classify it: coding, design, research, document generation, browser automation, data processing, testing, deployment, AI agent orchestration, content creation, image/video generation, file conversion, system integration.
3. Decide whether a reusable skill/tool may exist.
4. Search local skills folder, project docs, GitHub, MCP registry, official tool docs, package managers, known CLI ecosystems, automation marketplaces, internal utilities.
5. Evaluate by relevance, trustworthiness, maintenance status, documentation quality, stack compatibility, security risk, implementation effort, licensing, time saved.
6. Present a concise recommendation.

Skill Discovery Pass — when to run

Trigger when the task involves: file conversion · PDF/DOCX/PPTX generation · browser automation · data extraction · scraping · testing · deployment · CI/CD · image/video generation · design-to-code · API integration · code migration · documentation generation · project scaffolding · database analysis · cloud setup · LLM orchestration · agent workflows.

Skip when:

• the task is trivial
• the user explicitly asks for manual implementation
• searching would take longer than doing the task
• the project already has a known internal implementation path

Response Format

One good candidate

A reusable skill/tool looks like a good fit for this.

Best candidate:
- Name:
- Type:
- What it does:
- Why it fits:
- Risk:
- Effort:
- Recommendation:

Use this?
1. Yes, use it
2. No, build manually
3. Show alternatives

Multiple candidates (shortlist of three)

Three viable options:

1. [Tool/Skill Name]
   - Best for:
   - Pros:
   - Cons:
2. [Tool/Skill Name]
   - Best for:
   - Pros:
   - Cons:
3. [Tool/Skill Name]
   - Best for:
   - Pros:
   - Cons:

Recommendation: use [X], because [reason].

No good skill/tool exists

No reliable existing tool matches this task. Custom implementation is the better path.

Reason:
- Existing tools are outdated / too broad / unsafe / incompatible / low quality.

Next step:
Implement manually with a clean, minimal approach.

Important Behavior

• Do not blindly recommend tools.
• Do not choose the most popular tool if it does not fit.
• Prefer official tools and well-maintained repositories.
• Avoid abandoned GitHub projects unless there is no better option.
• Warn about security risks.
• Never install or run unknown tools without user approval.
• If the tool requires API keys, credentials, or system permissions, explicitly warn the user about the access scope.
• If the task is small and a tool would add unnecessary complexity, say so.

Approval Gate

Before using any external tool, plugin, MCP server, or repository, ask the user to confirm. Only proceed after approval.

Security Rules

• Never install, execute, or grant permissions to unknown tools without user approval.
• Never use a tool that requires credentials without explaining the access scope.
• Never prefer convenience over safety.

Goal

Prevent wasted implementation effort and tool chaos. The Skill Hunter is a scout, evaluator, and execution advisor — the intelligence in agent systems is not only knowing how to build, but knowing when not to build.