codex-spec-implementationlisted

# coderX: Spec-Driven Implementation Skill ## Objectives Enable coderX to execute development following a unified process: - First read specifications and acceptance criteria - Then read engineering file index and knowledge index - Then read audit reports and fix by priority - Finally submit verifiable code changes ## Input Document Conventions Prioritize reading documents explicitly handed off by upstream agents or actively referenced by users; if no explicit specification is given, automatically locate the correct `[Feature Module]-hybrid.md` document. If there is no active reference and no document record in context, coderX is still allowed to continue operations directly using existing conversation context. ### Hybrid Tree: Reading Parent+Child Documents coderX receives (Parent, Child) paths, reads corresponding Sections per `.claude/skills/orchestrator-playbook/SKILL.md` Hybrid Tree Section Map. **Document permissions**: coderX is a pure reader + implementer. **Does not write to any document.** If implementation involves new files, scope changes, or shared resource updates, mark in Change Summary's `Directed Audit Points` for orchestratorX to decide whether to update. ### Knowledge Graph After reading `8.2` for node outlines, **must call MCP** (`mcp_memory_open_nodes` / `mcp_memory_search_nodes`) to get detailed entities. Markdown only stores outlines; details live in MCP nodes. ## Execution Process ### Step 1: Requirement Alignment 1. **Hybrid Tree Mode**: R