What problem does it solve? Exploring a large codebase to gather context for an AI Orchestrator wastes tokens and dilutes signal when full files are read indiscriminately. This Skill enforces a disciplined discovery workflow that gathers only the structural context the Orchestrator needs. ## Core Features & Use Cases - Signature-First Exploration: Uses tree-sitter AST extraction (via custom_context_extract_signatures) to map classes, functions, interfaces, and types across Python, JavaScript, TypeScript, Java, Kotlin, Go, and Rust before reading any file bodies. - Vertical Slicing: Scopes discovery to a specific feature module (e.g., packages/billing/) while always injecting core SOP files like AGENTS.md and DESIGN.md into the report. - Strict Handover Guardrail: Compiles context reports into context-reports/ and halts, preventing the Executor from analyzing the reports itself. - Use Case: When the Orchestrator requests context for the authentication feature, run directory tree and signature extraction on src/features/auth/ only, then compile a report for upload instead of scanning the whole repository. ## Quick Start Ask the agent to follow the code-search skill to map the project structure and compile a context report for the Orchestrator.
