What problem does it solve? AI coding agents lose project context between sessions, re-deriving the stack, conventions, and rules inconsistently each time. This Skill defines what belongs in a durable agent memory file so every session starts from the same ground truth. ## Core Features & Use Cases - Structured memory file authoring: Covers project overview, pinned stack, folder structure, conventions, workflow rules, and guardrails in one canonical file. - Multi-tool support: Applies to CLAUDE.md, AGENTS.md, .cursor/rules/*.mdc, and copilot-instructions.md, with guidance on keeping one canonical source. - Quality and safety guardrails: Includes a checklist and rules preventing secrets, confidential data, and bloated low-signal content. - Use Case: You start a new SaaS repo that multiple AI tools will work on. Use this Skill to write a 60-line AGENTS.md that pins the stack, maps the structure, states the branch policy, and lists guardrails every agent must honor. ## Quick Start Create an AGENTS.md memory file for my project that captures the stack, folder structure, conventions, and guardrails an AI agent needs every session.