MEMORY.md — Manager Long-Term Memory

Company Context

Atomizer Engineering Co. is an AI-powered FEA optimization company.

Antoine runs NX/Simcenter on Windows (dalidou)
Optimization loop: agents prepare → Syncthing delivers → Antoine runs run_optimization.py → results flow back
All deliverables need Antoine's approval before going external
Quality over speed, but ship regularly

All Atomizer HQ planning docs are in context-docs/:

00-PROJECT-PLAN.md — The full project plan (vision, phases, success criteria)
01-AGENT-ROSTER.md — All 13 agents with detailed roles, capabilities, models
02-ARCHITECTURE.md — Technical architecture (Clawdbot multi-agent, Slack, Syncthing bridge)
03-ROADMAP.md — Phased rollout: Phase 0 (Core) → Phase 1 (Optimization) → Phase 2 (Production) → Phase 3 (Advanced)
04-DECISION-LOG.md — Key decisions: Clawdbot over Agent Zero, dedicated Slack workspace, phased rollout, autonomy with approval gates
05-FULL-SYSTEM-PLAN.md — Complete system specification (83KB, comprehensive)
README-ANTOINE.md — CEO's overview, the "why" behind everything Read these on first session to fully understand the vision and architecture.

Hydrotech Beam — Channel: #project-hydrotech-beam | Phase: DOE Phase 1 complete (39/51 solved, mass NaN fixed via commit 580ed65, displacement constraint relaxed 10→20mm). Next: pull fix on dalidou, rerun DOE.
Atomizer Project Standard — PKM: 2-Projects/P-Atomizer-Project-Standard/ | Phase: Specification complete (2026-02-18), awaiting CEO review. Deliverable: 00-SPECIFICATION.md (~57KB). Defines standardized folder structure, KB architecture, study lifecycle, naming conventions. Key decisions: numbered folders, unified KB, PROJECT.md + AGENT.md entry points, append-only DECISIONS.md. Next: Antoine review → iterate → build template → pilot on Hydrotech.

OP_11 — Digestion Protocol (CEO-approved 2026-02-11): STORE → DISCARD → SORT → REPAIR → EVOLVE → SELF-DOCUMENT. Runs at phase completion, weekly heartbeat, and project close. Antoine's corrections are ground truth.

Mass confusion (11.33 vs 1133 kg) — contradictions propagate fast when not caught. Digestion protocol's DISCARD + REPAIR phases exist to prevent this.
beam_lenght typo in NX — must use exact spelling. Domain-level knowledge.
NX integer expressions need unit=Constant, not MilliMeter
Always .resolve() paths, never .absolute() — NX file references break on copy
Existing optimization_engine should be wrapped, not reinvented
Sub-agents hit 200K token limits easily — keep prompts lean, scope narrow
Spawned sub-agents can't post to Slack channels (channel routing issue) — do Slack posting from main agent