Skip to content

meow:scout

Parallel codebase exploration that divides the project into segments and searches them simultaneously using Explore subagents.

What This Skill Does

meow:scout solves the "where is everything?" problem that comes up at the start of any complex task. It divides your codebase into logical segments (source, tests, config, types), spawns 2-6 parallel Explore agents to search them simultaneously, and returns a consolidated report with an architecture fingerprint, file map, complexity estimates, and routing suggestions for which agent should handle which area.

Core Capabilities

  • 3-tier search scope — Tier 1 (always: src, lib, config), Tier 2 (on request: tests, docs, migrations), Tier 3 (never: node_modules, .git, dist)
  • SCALE formula — Calculates optimal number of parallel agents: min(directory_count, 6)
  • Architecture fingerprint — Detects framework, language, patterns, monorepo status, test framework in 5 lines
  • Entry point identification — Surfaces main.ts, index.ts, app.py — the highest-signal files
  • Complexity estimation — File count + line count per area → low/medium/high
  • Handoff routing — Recommends which MeowKit agent should work on each area

When to Use This

Use meow:scout when...

  • Starting work on a feature that spans multiple directories
  • You need to understand project structure before planning
  • Debugging and need to find related files
  • You're new to a codebase and need orientation

Usage

bash
# Search for auth-related files
/meow:scout authentication

# Find database migration files
/meow:scout database migrations

# Understand project structure
/meow:scout project structure

Example Prompts

PromptWhat scout finds
/meow:scout authenticationAuth middleware, guards, login pages, auth tests, token utils
/meow:scout databaseModels, migrations, seeds, schema files, DB config
/meow:scout paymentPayment controllers, Stripe integration, billing tests

Quick Workflow

Query → SCALE Calculation → Tier Filtering
  → Parallel Explore Agents (2-6)
  → Deduplicate + Merge Results
  → Architecture Fingerprint + Entry Points
  → Complexity Estimates + Routing Suggestions
  → Scout Report (~2000 tokens max)

Skill Details

Phase: 0
Used by: orchestrator agent

Gotchas

  • Subagents returning partial results: Context window exceeded, agent returns truncated output → Set explicit file count limits per subagent; merge results with dedup
  • Missing hidden files in directory scan: Default glob patterns skip dotfiles → Include dotfiles explicitly when scanning config directories

Released under the MIT License.