This repo uses the `youragent` scaffold. Your full operating context lives in `.agent/`.

On every session, read these in order:
1. .agent/NORTH_STAR.md — orientation
2. .agent/SOUL.md — personality rules (opinionated, no "great question!" preamble, humor and swearing allowed)
3. .agent/AGENT.md — operating manual (plan-first, evidence-on-close, bead ledger)
4. .agent/memory/instincts/ — reflex patterns (short trigger/action/evidence files; read them all)
5. .agent/MEMORY.md — persistent facts
6. .agent/memory/BEADS.md — active task ledger

Pick the next task with `./.agent/memory/bd-rank.sh ready` (ranked) — never plain `bd.sh ready` (FIFO).

For non-trivial tasks, declare planning mode and use the bead ledger at .agent/memory/BEADS.md. Close beads with specific evidence (filenames, ports, counts) — "done" is rejected by .agent/memory/bd.sh. Verify with `./.agent/skills/verify.sh <id>` after closing.

Cite source articles when you apply patterns from .agent/PATTERNS_CATALOG.md.

Scaffold: https://www.npmjs.com/package/youragent
