README.md

Documentation Map

This directory is the maintainer-facing documentation for Agent Note. The public user guide lives in website/; this directory keeps implementation rules, architecture context, design notes, investigation history, and reusable assets close to the code.

Start Here

• engineering.md — implementation guidelines for humans and AI coding agents.
• architecture.md — current system architecture and data flow.
• knowledge/README.md — focused design notes, investigation history, and archived research.

Directory Layout

docs/
├── README.md
├── engineering.md
├── architecture.md
├── assets/
└── knowledge/
    ├── README.md
    ├── prompt-context.md
    ├── prompt-selection.md
    ├── agent-skill.md
    ├── agent-support-policy.md
    ├── investigations.md
    ├── research/
    └── archive/

The published Agent Skill itself lives outside docs/ at skills/agent-note/SKILL.md; the rationale and maintenance notes live in knowledge/agent-skill.md.

Reading Order

1. For code changes, read engineering.md first.
2. For schema, storage, action, dashboard, or agent-adapter changes, read architecture.md.
3. For prompt rendering or prompt filtering changes, read the matching file under knowledge/.
4. For old decisions, regressions, and historical plans, use knowledge/investigations.md and knowledge/archive/.

What Belongs Here

• Current implementation rules that should guide future code.
• Current architecture that must stay aligned with code.
• Design notes that explain non-obvious heuristics or storage decisions.
• Investigation history that helps prevent regressions.

What Does Not Belong Here

• End-user onboarding. Put that in website/ and the localized READMEs.
• Runtime instructions for a specific AI agent. Keep those in AGENTS.md / CLAUDE.md.
• Contributor workflow basics. Keep those in CONTRIBUTING.md.
• Temporary scratch notes. Move resolved material into knowledge/investigations.md or delete it.