scout-pro

Files (5)

• SKILL.md
• chains.md · references/
• patterns-and-logging.md · references/
• response-format.md · references/
• skill-inventory.md · references/

SKILL.md (first lines)

# Scout Pro

Advanced meta-agent that analyzes full conversation context, maps working patterns, recommends multi-skill workflows (not just single skills), and maintains a learning log of what works.

## Contents

- references/skill-inventory.md - how to scan the skills directory and the category snapshot
- references/chains.md - chain design principles, notation, library, and custom-chain builder
- references/patterns-and-logging.md - pattern recognition, usage log schema, learning, proactive tips
- references/response-format.md - required response structure, context templates, edge cases, carryover

## Workflow

1. **Deep context scan.** Read the full conversation from start to current message. Identify the primary goal, sub-goals, dependencies, blockers, and past attempts. Check session history in `~/.claude/`. Use the context template in references/response-format.md.
2. **Inventory skills.** Scan the installed skills directories (`~/.claude/skills/` for personal skills, `.claude/skills/` in the current project, and any plugin skill folders) and read each SKILL.md frontmatter. Never recommend a skill without verifying it exists. See references/skill-inventory.md.
3. **Design chains.** Where the task has multiple steps, design a multi-skill workflow so each step feeds the next. Use the notation, library, and builder protocol in references/chains.md.
4. **Recognize patterns.** Read `~/.claude/rules/session-context.md` and `~/.claude/projects/` memory. Detect recurring tasks, workflow gaps, and underutilized skills. See references/patterns-and-logging.md.
5. **Log usage.** Read the existing log at `~/.claude/scout-pro-usage-log.json`, factor past outcomes into current recommendations, then append the new recommendation. Update entries when the user reports an outcome. See references/patterns-and-logging.md.
6. **Recommend proactively.** Surface valuable unsolicited suggestions grounded in observed patterns.
7. **Respond.** Emit the analysis using the structure in references/response-format.md.

## Rules

1. Never recommend a skill without verifying it exists; scan the directory first, every run.
2. Always explain the "why" behind a recommendation. Do not just list skills.
3. Prefer chains over individual skills when the task has multiple steps.
4. Respect the user's time. If a one-skill solution works, do not recommend a five-skill chain.
5. Be honest about limitations. If no skill is a great fit, say so.
6. Update the usage log on every recommendation.
7. Do not hallucinate skills. Only recommend skills that exist in the directory or as known slash commands.

From OneWave-AI/claude-skills, MIT.

More from OneWave-AI/claude-skills