AGENTS.md — Agent-Agnostic Instructions

This file provides operating instructions for any AI agent working in this vault — whether that's Claude Code, OpenAI Codex, OpenCode, Cursor, GitHub Copilot, or any other agentic harness.

Read This First

All operating rules live in /schema/agent-protocol.md. That file is the single source of truth. This file adds agent-agnostic guidance only.

Vault Architecture

This is a four-zone Second Brain vault:

Core Workflows

Ingest a Source

1. Read the source file in /sources/.
2. Update its frontmatter status to reading.
3. Create or update wiki pages in /wiki/ that synthesize the source content.
4. Use [source-filename](/wiki/source-filename) citations in wiki pages.
5. Update the source frontmatter status to processed.

Research a Topic

1. Search existing wiki pages and sources for relevant content.
2. If gaps exist, the human provides new sources to /sources/.
3. Ingest new sources (see above).
4. Create a wiki page synthesizing findings.

Daily Journal

1. Create a file at /journal/YYYY-MM-DD.md using the journal template.
2. The human fills in the content. You read it for context.

Vault Status

1. Scan /sources/ for status: unread files.
2. Scan /wiki/ for status: draft pages that haven't been updated recently.
3. Check /wiki/contradictions.md for unresolved conflicts.
4. Report findings to the human.

Key Rules

1. Never modify reviewed or stable wiki pages without explicit instruction.
2. Never modify files in /sources/.
3. Always include frontmatter — see /schema/agent-protocol.md for formats.
4. Flag contradictions in /wiki/contradictions.md.
5. No fabrication. If the source doesn't say it, don't write it.
6. One concept per wiki page. Split broad pages.

Agent-Specific Files

Different agents use different instruction files. All point to the same source of truth: