What problem does it solve? AI coding agents produce poor output when they receive too little context (hallucinated APIs, ignored conventions) or too much (lost focus, degraded quality). This Skill provides a structured method for deciding what information an agent sees, when it sees it, and how it is organized across a session. ## Core Features & Use Cases - Five-Level Context Hierarchy: Structures context from persistent rules files (CLAUDE.md, .cursorrules, AGENTS.md) down through specs, source files, error output, and conversation history. - Digest Delegation Pattern: Delegates heavy codebase reading to subagents that return capped, structured summaries, keeping the main session's context window free for planning. - Confusion Management: Provides explicit patterns for surfacing spec conflicts, missing requirements, and ambiguity instead of silently guessing. - Use Case: When starting a new feature, you load only the relevant spec section, the files to modify, one existing pattern example, and a rules file — then verify the agent references real project APIs rather than invented ones. ## Quick Start Ask the agent to help you write a CLAUDE.md rules file covering your tech stack, build commands, conventions, and boundaries before starting the next coding task.
