Menu
Fast codebase scouting using parallel agents. Use for file discovery, task context gathering, searching across directories.
Install with Codex or Claude Copy this prompt, paste it into Codex, Claude, or another assistant, and let it review the skill page and install it for you.
Based on SOC occupation classification
Search library/framework documentation via llms.txt (context7.com). Use for API docs, GitHub repository analysis, technical documentation lookup, latest library features.
ALWAYS activate before implementing ANY feature, plan, or fix.
Debug systematically with root cause analysis before fixes. Use for bugs, test failures, unexpected behavior, performance issues, call stack tracing, multi-layer validation, log analysis, CI/CD failures, database diagnostics, system investigation.
ALWAYS activate before fixing ANY bug, error, test failure, CI/CD issue, type error, lint, log error, UI issue, code problem.
Unified knowledge retrieval skill. Build traceable answers from git, MCP, and memory files.
Plan implementations, design architectures, create technical roadmaps with detailed phases.
| name | hi:scout |
| description | Fast codebase scouting using parallel agents. Use for file discovery, task context gathering, searching across directories. |
| argument-hint | [search-target] [ext] |
| metadata | {"author":"baka3k","version":"2.0.0"} |
Fast, token-efficient codebase scouting using parallel agents.
Parse user prompt for targets, key directories, patterns, file types. Determine SCALE (subagent count).
Split codebase logically. Each agent gets distinct scope, no overlap.
Skip if <=2 agents or Task tools unavailable. TaskList -> use existing. Else TaskCreate per agent with scope metadata. Metadata: agentType, scope, scale, agentIndex, totalAgents, toolMode, priority:P2, effort:3m. See references/task-management.md for patterns.
Load internal (references/internal-scouting.md) or external (references/external-scouting.md). TaskUpdate each to in_progress before spawning. Max 3m per agent. Skip timed-out.
Aggregate: deduplicate paths, merge descriptions, note gaps/timeouts, list unresolved.
# Scout Report
## Relevant Files
- path/to/file.ts - description
## Unresolved Questions
- any gaps
Read .claude/.ck.json: gemini.model (default: gemini-3-flash-preview).
Invoke /hi:project-organization to organize outputs.