feat: add Atomizer HQ multi-agent cluster infrastructure

- 8-agent OpenClaw cluster (Manager, Tech-Lead, Secretary, Auditor, Optimizer, Study-Builder, NX-Expert, Webster) - Orchestration engine: orchestrate.py (sync delegation + handoffs) - Workflow engine: YAML-defined multi-step pipelines - Agent workspaces: SOUL.md, AGENTS.md, MEMORY.md per agent - Shared skills: delegate, orchestrate, atomizer-protocols - Capability registry (AGENTS_REGISTRY.json) - Cluster management: cluster.sh, systemd template - All secrets replaced with env var references
2026-02-15 21:18:18 +00:00
parent d6a1d6eee1
commit 3289a76e19
170 changed files with 24949 additions and 0 deletions
--- a/hq/workspaces/manager/memory/2026-02-08.md
+++ b/hq/workspaces/manager/memory/2026-02-08.md
@@ -0,0 +1,51 @@
+# 2026-02-08
+
+## New Project: Hydrotech Beam Structural Optimization
+- **Channel:** #project-hydrotech-beam
+- **Request:** Optimize I-beam with sandwich cross-section — reduce mass, reduce tip displacement, keep stress safe
+- **Status:** Intake received, project folder created at `/home/papa/atomizer/projects/hydrotech-beam/`
+- **Next:** Delegate technical breakdown to Technical Lead (OP_09 → OP_10 Step 2)
+- 4 design variables, NX Nastran static analysis, steel beam with lightening holes
+- Current baseline: ~974 kg, ~22 mm displacement
+- Targets: minimize mass, displacement < 10 mm, stress < 130 MPa
+
+## Issues Raised by Antoine
+- No Notion project page yet (Phase 2 feature — not available)
+- Secretary had no project context — briefed her via sessions_send, she's now tracking it
+- Slowness noted — all agents on Opus, expected for now
+- Project folder confirmed at `/home/papa/atomizer/projects/hydrotech-beam/`
+- Antoine wants cross-agent context sharing to work better — need to think about how Secretary gets project updates automatically
+
+## Config Changes
+- Added `#project-hydrotech-beam` (C0AE4CESCC9) to channel config with `requireMention: false`
+- Antoine no longer needs to tag to get a response in project channels
+
+## Antoine Request: Project Dashboard & File Access
+- Manager now owns ALL admin responsibility — Mario only for infrastructure bridges
+- NO Notion — Antoine doesn't use it
+- Project data should live in Atomizer repo (Gitea: http://100.80.199.40:3000/Antoine/Atomizer.git)
+- Documentation = efficient .md files in the repo
+- Current project files at `/home/papa/atomizer/projects/hydrotech-beam/` need to move into `/home/papa/repos/Atomizer/projects/hydrotech-beam/`
+- Syncthing syncs: ATODrive, Atomaste, obsidian-vault, Sync — NOT the atomizer workspace
+- Atomizer repo is git-managed (correct approach for project data)
+
+## Project Structure Overhaul — COMPLETED
+- Designed and implemented KB-integrated project structure for Atomizer
+- Hydrotech Beam restructured: README, CONTEXT, BREAKDOWN, DECISIONS, models/, kb/, studies/, deliverables/
+- KB initialized: components/sandwich-beam.md, materials/steel-aisi.md, fea/models/sol101-static.md
+- Gen 001 created from intake + technical breakdown
+- 6 decisions logged in DECISIONS.md (DEC-HB-001 through DEC-HB-006)
+- Created `knowledge-base-atomizer-ext.md` — Atomizer extension of Mario's shared KB skill
+- Extension pattern: use base skill as-is, extend with Atomizer-specific agent workflows
+- All committed to Gitea: commit 9541958
+- Channel config fixed: #project-hydrotech-beam no longer requires mention
+
+## Repo Cleanup — IN PROGRESS
+- CEO approved major docs cleanup of Atomizer repo
+- Spawned sub-agent (label: repo-cleanup) to handle:
+  - Archive stale docs (RALPH_LOOP, old CANVAS plans, dashboard iterations) to docs/archive/review/
+  - Create docs/hq/ with agent-facing documentation (PROJECT_STRUCTURE, KB_CONVENTIONS, AGENT_WORKFLOWS, STUDY_CONVENTIONS)
+  - Update docs/00_INDEX.md
+- KB skill discussion resolved: Mario's pipeline = tool, Atomizer owns project KB, no duplication
+- Mario's KB output can bootstrap Atomizer project KB if available
+- CAD-Documenter tool being renamed (to KBS or similar) — update references when it lands
--- a/hq/workspaces/manager/memory/2026-02-09.md
+++ b/hq/workspaces/manager/memory/2026-02-09.md
@@ -0,0 +1,61 @@
+# 2026-02-09
+
+## Phase 1 Approved & Kicked Off
+- Antoine asked if Phase 0 was complete enough to move forward
+- Assessed Phase 0 at ~75% complete (structure proven, execution loop not yet tested)
+- Recommended proceeding to Phase 1 with Hydrotech Beam as the validation project
+- Antoine approved — Phase 1 is live
+
+## Phase 1 Agents — Status
+- Optimizer, Study Builder, Auditor are NOT yet configured as gateway agents
+- Only Manager, Secretary, Technical Lead exist as real agents
+- Using sessions_spawn sub-agents as a workaround for now
+- Need Mario to set up the actual agent workspaces + gateway config for Phase 1 agents
+
+## Hydrotech Beam — Resuming
+- Posted project kickoff in #project-hydrotech-beam with full assignment roster
+- Workflow: serial chain managed by me (per DEC-A003)
+  - Step 1: Optimizer designs strategy ← IN PROGRESS (spawned sub-agent)
+  - Step 2: Auditor reviews plan
+  - Step 3: Study Builder writes code
+  - Step 4: Auditor reviews code
+  - Step 5: CEO approves for execution
+  - Step 6: Run on Windows (manual)
+  - Step 7: Results analysis
+- 9 technical gaps still open from Tech Lead's breakdown (G1-G9)
+- Optimizer working from BREAKDOWN.md to produce OPTIMIZATION_STRATEGY.md
+
+## Antoine Questions
+- Asked about workflow management (serial vs parallel) — explained I manage the chain
+- Asked about roll call location — posted project kickoff in #project-hydrotech-beam
+- Asked "what's next? Where do I review?" — gave full status briefing in #all-atomizer-hq
+  - Pointed to Gitea as the browsable dashboard
+  - Recommended resolving 9 gaps as top priority
+  - Proposed: daily auto-status from Secretary, README as live dashboard
+  - Antoine wants proactive improvement — gave 6 prioritized recommendations
+
+## File Access Gap Identified
+- Atomizer repo NOT synced to Windows (dalidou) via Syncthing
+- Only ATODrive, Atomaste, obsidian-vault, Sync are shared
+- Model files (Beam.prt, etc.) never added to models/ — placeholder only
+- Antoine can't browse KB or project docs from Windows
+- **Resolution:** Antoine setting up Syncthing for `projects/hydrotech-beam/` specifically
+  - Server path: `/home/papa/repos/Atomizer/projects/hydrotech-beam/`
+  - Rest of repo stays git-only (he has Gitea web access from Windows)
+  - .gitignore allows .prt/.fem/.sim (only .bak excluded)
+  - Once sync is live, model files land in models/ and I commit to Gitea
+- Antoine wants KBS session but needs model files accessible first
+
+## Single Source of Truth — Consolidation Done
+- **Canonical project path:** `/home/papa/repos/Atomizer/projects/hydrotech-beam/` (Gitea + Syncthing)
+- Removed stale duplicate at `/home/papa/atomizer/projects/hydrotech-beam/`
+- Created symlink so old references still resolve
+- Cleaned up Syncthing conflict files
+- All agents should reference `/repos/Atomizer/projects/` from now on
+- Antoine dropping remaining model files via Syncthing from Windows
+
+## Improvement Initiatives (Self-Directed)
+- [ ] Set up Secretary daily status posts
+- [ ] Update Hydrotech README to be a live status card
+- [ ] Track gap resolution progress
+- [x] Consolidate project folder to single source of truth (repo)
--- a/hq/workspaces/manager/memory/2026-02-10.md
+++ b/hq/workspaces/manager/memory/2026-02-10.md
@@ -0,0 +1,135 @@
+# 2026-02-10
+
+## Hydrotech Beam — KBS Sessions Received
+- Antoine recorded 3 KBS capture sessions on his Windows machine (NX/Simcenter)
+- Data location: `/home/papa/ATODrive/Projects/hydrotech-beam/Hydrotech-Beam/_capture/`
+- Sessions: `20260210-132817` (6s), `20260210-161401` (38s), `20260210-163801` (414s main session)
+- Main session is a full walkthrough of the NX model with parameter names, values, BCs, materials
+
+### New Information from KBS Sessions
+- Beam length = 5,000 mm (`beam_length` expression)
+- Cantilever: left fixed, right loaded with 10,000 kgf downward
+- Hole span = 4,000 mm (`p6`), holes start/end 500mm from beam ends
+- Mass via expression `p1` (NOT `p173` as we had) — starting 11.33 kg (CONTRADICTS 974 kg baseline!)
+- Material: ANSI Steel 1005 — future: aluminum 6061, stainless ANSI 310
+- Mesh: CQUAD4 thin shells, mid-surface idealization, element size = 67.4/2
+- New expression names: `beam_half_height`, `beam_half_width`
+- `p6` (hole span) as potential new design variable
+- 4 screenshot triggers in the session metadata
+
+### Actions Taken
+- Posted acknowledgment + next steps in #project-hydrotech-beam
+- Spawned Tech Lead sub-agent (label: tech-lead-kb-update) to:
+  - Process all 3 transcripts
+  - Update KB to Gen 002
+  - Reconcile mass discrepancy (11.33 kg vs 974 kg)
+  - Close resolved gaps (G1, G2, G5 partial, G8)
+  - Update CONTEXT.md
+  - Commit to Gitea
+
+### Workflow Status
+- Step 1 (Optimizer strategy): OPTIMIZATION_STRATEGY.md exists as DRAFT from Feb 9
+- Current: Processing new KB data before proceeding
+- Next: Optimizer revises strategy with confirmed params → Auditor review → Study Builder code
+- Model files confirmed synced: Beam.prt, Beam_fem1.fem, Beam_fem1_i.prt, Beam_sim1.sim
+
+### Completed
+- [x] Tech Lead completed KB Gen 002 update — commit `b88657b`
+- [x] Mass corrected AGAIN: **1,133.01 kg** (`p173`), NOT 11.33 kg — Antoine corrected us
+- [x] Binary introspection of Beam.prt — extracted complete expression table (commit `15a457d`)
+- [x] DV baselines are NOT round: face_thickness=21.504, core_thickness=25.162 (not 20/20)
+- [x] Gaps G12-G14 closed (beam_half_height=250, beam_half_width=150, holes_diameter expression confirmed)
+- [x] Important: `beam_lenght` has TYPO in NX (no 'h') — scripts must use exact spelling
+- [x] `hole_count` links to `Pattern_p7` in the NX pattern feature
+- [x] CONTEXT.md updated with full expression map, pushed to Gitea
+
+### Pending — Waiting on Antoine
+- [ ] Baseline re-run (G10, G11) — need current displacement and stress values
+- [x] Decision on `p6` (hole span) — kept fixed at 4,000mm for now (Manager decision)
+
+### Windows Environment (dalidou)
+- Path: `C:\Users\antoi\Atomizer\projects\hydrotech-beam\` (Syncthing from server)
+- Python: `anaconda3\envs\atomizer` (conda env named "atomizer")
+- Antoine ran smoke test on Feb 11 — hit 2 bugs, both fixed (commit `135698d`)
+- NXOpen implementation still needed (solve, extract_displacement, extract_stress)
+
+### In Progress
+- [x] Optimization strategy updated with corrected baselines (commit `3e51804`)
+- [x] Auditor review: APPROVED WITH CONDITIONS — 2 blockers found and fixed:
+  - Hole spacing formula: `span/(n-1)` not `span/(n+1)` — fixed
+  - Web height constraint: added `500 - 2*face - dia > 0` pre-check — fixed
+  - Commit `94bff37`
+- [x] Study Builder completed Phase 1 code (commit `017b90f`) — verified end-to-end with stub solver
+  - 6 files: run_doe.py, sampling.py, geometric_checks.py, nx_interface.py, requirements.txt, README.md
+  - Pre-flight geometric filter catches ~24% of infeasible combos
+  - NXOpen template ready — needs 3 methods filled in on Windows (solve, extract_disp, extract_stress)
+- [ ] Antoine running baseline SOL 101 for displacement + stress (parallel)
+- [ ] `p6` kept fixed at 4,000mm for now (DEC by Manager)
+
+### NXOpenSolver → Existing Engine Integration (Late Evening)
+- Antoine confirmed: runs everything from his "Honda atomizer" conda env on Windows
+- Uses existing `run_optimization.py` which calls `NXSolver` + `NXParameterUpdater` + pyNastran extractors
+- **Key insight:** We do NOT need to write NXOpen code from scratch — the Atomizer engine already has everything:
+  - `optimization_engine/nx/solver.py` — journal-based solver via `run_journal.exe`
+  - `optimization_engine/nx/updater.py` — expression updates via `.exp` import
+  - `optimization_engine/extractors/extract_displacement.py` — pyNastran OP2
+  - `optimization_engine/extractors/extract_von_mises_stress.py` — pyNastran OP2, kPa→MPa
+  - `optimization_engine/extractors/extract_mass_from_expression.py` — from temp file
+- Delegated to Study Builder (label: `study-builder-nx-impl`) to rewrite `NXOpenSolver` as a wrapper around existing engine
+- Asked Antoine to confirm `pyNastran` is installed in the conda env
+
+### Infra Fixes
+- Study Builder model was set to non-existent `claude-sonnet-5` → fixed to `claude-sonnet-4-20250514`
+- All agents were missing Anthropic API auth → propagated from Manager's auth-profiles.json
+- Agents fixed: secretary, study-builder, optimizer, auditor, technical-lead
+
+### Study Builder Delivered — NXOpenSolver (commit `33180d6`)
+- Wraps existing Atomizer engine: NXParameterUpdater, NXSolver, pyNastran extractors
+- HEEDS-style iteration folders, 600s timeout, CQUAD4 shell stress, kPa→MPa
+- Full interface compatibility with run_doe.py preserved
+- 252 additions, 126 deletions
+
+### Tech Lead Refined — NXOpenSolver v2 (commit `390ffed`)
+- Built on Study Builder's work with improvements:
+  - Element type auto-detection (tries solids first, falls back to CQUAD4)
+  - OP2 fallback path (solver result → expected naming convention)
+  - Mass fallback via `_temp_part_properties.json`
+  - Follows SAT3_Trajectory_V7 FEARunner pattern exactly
+- Both commits stack cleanly on main, latest is the active version
+
+### Late Night — Antoine Follow-Up (~23:00-01:00 UTC)
+- Antoine returned: "Yeah! What's next?" — confirmed ready to move forward
+- Asked about conda env: confirmed he uses `conda atomizer` (defined in `environment.yml` at repo root)
+  - Includes optuna, scipy, numpy, pandas, pyNastran — all Phase 1 deps covered
+- Asked "What's the NXOpen implementation about?" — explained the 3 bridge methods (solve, extract_disp, extract_stress)
+- Antoine asked how this relates to legacy Atomizer studies (SAT3, mirror blank)
+  - Confirmed: same engine (NXSolver, NXParameterUpdater, pyNastran extractors)
+  - Differences: geometric pre-filter, LHS sampling, cleaner separation, project-scoped
+- **Antoine approved:** "go ahead and do it"
+- Delegated NXOpen implementation completion to Technical Lead (label: `hydrotech-nxopen-impl`)
+  - Task: complete NXOpenSolver.evaluate() using existing Atomizer engine components
+  - Reference: SAT3_Trajectory_V7, bracket study, existing engine classes
+
+### Feb 11 Morning — Bug Fixes + Final Refactor
+- Antoine tested on dalidou, hit 2 bugs:
+  1. SQLite duplicate study name → fixed with `load_if_exists=True` + `--clean` flag
+  2. Sampling crash with `n-samples 1` → skip stratified patching when n < 11
+  - Commit `135698d`
+- **Full refactor of nx_interface.py** (commit `126f0bb`):
+  - `AtomizerNXSolver` wraps existing `optimization_engine` (NXSolver + pyNastran extractors)
+  - HEEDS-style iteration folders, .exp file generation, OP2 extraction
+  - StubSolver improved with beam-theory approximations
+  - Windows path confirmed: `C:\Users\antoi\Atomizer\projects\hydrotech-beam\`
+  - Conda env: `atomizer` (all deps pre-installed)
+
+### Future Initiative — NX Simcenter 3D MCP (CEO request, Feb 11)
+- MCP server on dalidou for direct NXOpen interaction
+- Endpoints: expressions.list/get/set, model.update, solve.run, results.*, introspect, screenshots
+- Eliminates pyNastran, temp files, journal generation — all via NXOpen API
+- Target: Phase 2/3 roadmap
+- Logged per Antoine's explicit request — not blocking current work
+
+### Next
+- [ ] Antoine tests `--backend nxopen` on dalidou (single trial smoke test)
+- [ ] Full 51-trial Phase 1 run
+- [ ] Phase 2 TPE optimization
--- a/hq/workspaces/manager/memory/2026-02-11.md
+++ b/hq/workspaces/manager/memory/2026-02-11.md
@@ -0,0 +1,29 @@
+# 2026-02-11
+
+## Channel Config
+- Added #research-and-development (C0AEB39CE5U) to Slack config
+- All 6 agents bound to it
+- Set `requireMention: false` globally for all Slack channels per Antoine's request
+
+## NXOpen MCP Server — INSTALLED ✅
+- **Repo**: `http://100.80.199.40:3000/Antoine/NXOpen-MCP.git`
+- **Local path**: `/home/papa/atomizer/tools/nxopen-mcp/`
+- **Venv**: `.venv/` with CPU-only torch (no CUDA needed)
+- **Data**: 203MB pre-indexed ChromaDB + JSON caches
+  - 15,219 NXOpen classes, 64,320 methods
+  - 149 nxopentse functions
+  - 287 pyNastran classes
+- **Run**: `./run-server.sh` or `python -m nxopen_mcp.server --data-dir ./data`
+- **Protocol**: stdio-based MCP
+- **Tools**: search_nxopen, get_class_info, get_method_info, get_examples, list_namespaces
+- Wired into NX Expert agent via exec/Python subprocess
+
+## NX Expert Agent — HIRED ✅
+- **Agent ID**: `nx-expert`
+- **Model**: Sonnet 4 (cost-effective specialist)
+- **Workspace**: `/home/papa/atomizer/workspaces/nx-expert/`
+- **Channels**: #hq (C0AEJV13TEU), #research-and-development (C0AEB39CE5U)
+- **Mention patterns**: @nx-expert, @NX Expert, @nx, 🖥️
+- **Tools**: NXOpen MCP via Python exec, atomizer-protocols skill
+- **Role**: NX Open API expert, solver config, element selection, journal scripting
+- First Phase 2 agent to come online — ahead of schedule
--- a/hq/workspaces/manager/memory/2026-02-13.md
+++ b/hq/workspaces/manager/memory/2026-02-13.md
@@ -0,0 +1,41 @@
+# 2026-02-13
+
+## Nightly Digestion Cron — LIVE ✅
+- **Job ID:** `e157faf0-084f-4d8d-8693-814cf4340a48`
+- **Schedule:** Every night at 4:00 AM ET (`0 4 * * *` America/Toronto)
+- **Type:** Isolated agentTurn (manager), announces to #all-atomizer-hq
+- **Protocol:** OP_11 full 6-step cycle (STORE → DISCARD → SORT → REPAIR → EVOLVE → SELF-DOCUMENT)
+- Set up per Antoine's directive to officialize nightly memory processing
+
+## Hydrotech Beam — Resumed
+- Antoine approved continuing to next phase (~01:36 UTC)
+- DOE Phase 1 (51 trials) completed previously but **gate check FAILED**:
+  - 39/51 solved, 12 geo-infeasible (hole overlap)
+  - **0 fully feasible designs** — displacement ≤10mm never achieved (min ~19.6mm)
+  - **Mass = NaN** on all trials — extraction bug in journal/script
+  - Stress constraint (≤130 MPa) met by some trials but displacement kills everything
+- **Delegated to Tech Lead:** Diagnose mass NaN, analyze DOE landscape, recommend feasibility fix
+  - Spawned sub-agent session: `hydrotech-doe-analysis`
+- **Pending CEO decision:** Relax 10mm displacement constraint? Options presented: relax to ~20mm, expand DVs, or keep and find boundary
+- Optimizer + Study Builder on standby for Phase 2 (TPE) after fixes
+
+## Mass NaN Fix — COMMITTED ✅
+- **Commit:** `580ed65` on Atomizer repo main branch
+- **Root cause:** `solve_simulation.py` journal's `solve_simple_workflow()` tried to read mass via expression `p173` after part switching (geom→FEM→SIM→solve→back). Expression was stale/inaccessible after switching. `_temp_mass.txt` never written.
+- **NOT** the `M1_Blank` hardcoding (that's assembly workflow only). Beam uses `solve_simple_workflow` (no `.afm`).
+- **Fix (2 edits):**
+  1. Extract mass RIGHT AFTER geometry rebuild (`DoUpdate()`) while geom part is work part — uses `MeasureManager.NewMassProperties()` (computes fresh from solid bodies)
+  2. Post-solve: skip re-extraction if already done; fallback to MeasureManager instead of p173
+- **NX Expert** did the fix but did NOT use MCP server — was a code-level debug task, not API discovery
+- **NX Expert Slack issue:** sub-agent couldn't post to #all-atomizer-hq (channel ID routing problem for spawned agents)
+- **Next:** Pull on dalidou, test single trial, then re-run full DOE with 20mm constraint
+
+## Sub-agent Issues
+- Tech Lead sub-agents both hit 200K token context limit and aborted (`abortedLastRun: true`)
+- Had to do diagnosis myself then delegate to NX Expert
+- NX Expert also couldn't post to Slack (channel_not_found with various target formats)
+- **Lesson:** Sub-agents need leaner prompts, and Slack channel routing needs fixing for spawned sessions
+
+## DEC-HB-012 — Displacement Constraint Relaxed
+- 10mm → 20mm, CEO approved (dummy case, pipeline proving)
+- Updated CONTEXT.md and DECISIONS.md in project folder
--- a/hq/workspaces/manager/memory/projects/dashboard-reporting-masterplan.md
+++ b/hq/workspaces/manager/memory/projects/dashboard-reporting-masterplan.md
@@ -0,0 +1,236 @@
+# 📊 Atomizer Dashboard & Reporting System — Master Plan
+
+> **Status:** PROPOSAL | **Date:** 2026-02-14 | **Author:** Manager Agent | **For:** Antoine (CEO)
+
+---
+
+## Executive Summary
+
+A file-based, agent-native dashboard and reporting system that gives Antoine real-time project visibility without leaving the existing Atomizer stack. No new infrastructure—just structured markdown, automated aggregation, and agent-generated reports.
+
+---
+
+## 1. Information Architecture
+
+```
+shared/
+├── PROJECT_STATUS.md          ← Single source of truth (Manager-owned)
+├── project_log.md             ← Append-only agent activity log
+├── dashboards/
+│   ├── exec-summary.md        ← CEO dashboard (auto-generated)
+│   ├── technical.md           ← FEA/optimization status
+│   └── operations.md          ← Agent health, queue, throughput
+├── reports/
+│   ├── weekly/                ← YYYY-WXX-report.md
+│   ├── project/               ← Per-project closeout reports
+│   └── templates/             ← Report templates (markdown)
+├── data-contracts/
+│   └── schemas.md             ← Field definitions for all status files
+└── kpi/
+    └── metrics.md             ← Rolling KPI tracker
+```
+
+**Principle:** Everything is markdown. Agents read/write natively. No database, no web server, no maintenance burden.
+
+---
+
+## 2. Dashboard Modules
+
+### 2A. Executive Summary (`dashboards/exec-summary.md`)
+**Audience:** Antoine | **Update frequency:** On every PROJECT_STATUS.md change
+
+| Section | Content |
+|---------|---------|
+| 🚦 Project RAG | Red/Amber/Green per active project, one line each |
+| 📌 Decisions Needed | Items blocked on CEO approval |
+| 💰 Resource Burn | Agent token usage / cost estimate (daily/weekly) |
+| 🏆 Wins This Week | Completed milestones, delivered studies |
+| ⚠️ Top 3 Risks | Highest-impact risks across all projects |
+
+**Format:** ≤30 lines. Scannable in 60 seconds.
+
+### 2B. Technical Dashboard (`dashboards/technical.md`)
+**Audience:** Technical Lead, Optimizer | **Update frequency:** Per study cycle
+
+| Section | Content |
+|---------|---------|
+| Active Studies | Study name, iteration count, best objective, convergence % |
+| FEA Queue | Jobs pending / running / completed / failed |
+| Model Registry | Active NX models, mesh stats, last validated date |
+| Optimization Curves | Tabular: iteration vs objective vs constraint satisfaction |
+| Knowledge Base Delta | New entries since last report |
+
+### 2C. Operations Dashboard (`dashboards/operations.md`)
+**Audience:** Manager (self-monitoring), Mario (infra) | **Update frequency:** Hourly via cron or on-demand
+
+| Section | Content |
+|---------|---------|
+| Agent Health | Last active timestamp per agent, error count (24h) |
+| Message Throughput | Messages processed per agent per day |
+| Queue Depth | Pending delegations, blocked tasks |
+| Token Budget | Usage vs budget per agent, projected monthly |
+| System Alerts | Disk, memory, process status flags |
+
+---
+
+## 3. Data Contracts
+
+Every agent writing to `project_log.md` MUST use this format:
+
+```markdown
+## [YYYY-MM-DD HH:MM] agent-id | project-slug | event-type
+
+**Status:** in-progress | completed | blocked | failed
+**Summary:** One-line description
+**Detail:** (optional) Multi-line context
+**Metrics:** (optional) key=value pairs
+**Blockers:** (optional) What's blocking and who can unblock
+
+---
+```
+
+### Event Types (enumerated)
+| Type | Meaning |
+|------|---------|
+| `study-start` | New optimization study launched |
+| `study-iteration` | Iteration completed with results |
+| `study-complete` | Study converged or terminated |
+| `review-request` | Deliverable ready for review |
+| `decision-needed` | CEO/human input required |
+| `task-delegated` | Work handed to another agent |
+| `task-completed` | Delegated work finished |
+| `error` | Something failed |
+| `milestone` | Phase/gate achieved |
+
+### Dashboard Field Schema
+Each dashboard section maps to specific log event types. Manager agent aggregates—no other agent touches dashboard files directly.
+
+---
+
+## 4. Report System
+
+### 4A. Weekly Report (auto-generated every Friday or on-demand)
+**Template:** `reports/templates/weekly-template.md`
+
+```markdown
+# Atomizer Weekly Report — YYYY-WXX
+
+## Highlights
+- (auto: completed milestones from log)
+
+## Projects
+### [Project Name]
+- Status: RAG
+- This week: (auto: summary of log entries)
+- Next week: (auto: from PROJECT_STATUS.md planned items)
+- Blockers: (auto: open blockers)
+
+## KPIs
+| Metric | This Week | Last Week | Trend |
+|--------|-----------|-----------|-------|
+
+## Agent Performance
+| Agent | Messages | Tasks Done | Errors | Avg Response |
+|-------|----------|------------|--------|-------------|
+
+## Decisions Log
+- (auto: from decision-needed events + resolutions)
+```
+
+### 4B. Project Closeout Report
+Generated when a project reaches `completed` status. Includes full decision trail, final results, lessons learned, KB entries created.
+
+### 4C. On-Demand Reports
+Antoine can request via Discord: "Give me a status report on [project]" → Manager generates from log + status files instantly.
+
+### 4D. PDF Generation
+Use existing `atomaste-reports` skill for client-facing PDF output when needed.
+
+---
+
+## 5. Documentation Governance
+
+### Two-Tier System
+
+| Tier | Location | Owner | Mutability |
+|------|----------|-------|------------|
+| **Foundational** | `context-docs/` | Mario + Antoine | Immutable by agents. Amended only via CEO-approved change request. |
+| **Project-Specific** | `shared/`, `memory/projects/` | Manager (gatekeeper), agents (contributors) | Living documents. Agents write, Manager curates. |
+
+### Rules
+1. **Foundational docs** (00-05, SOUL.md, protocols) = constitution. Agents reference, never edit.
+2. **Project docs** = operational. Agents append to log; Manager synthesizes into status files.
+3. **Dashboards** = derived. Auto-generated from project docs. Never manually edited.
+4. **Reports** = snapshots. Immutable once generated. Stored chronologically.
+5. **Knowledge Base** = accumulative. Grows per project via `cad_kb.py`. Never pruned without review.
+
+### Change Control
+- Protocol changes → Antoine approval → Mario implements → agents reload
+- Dashboard schema changes → Manager proposes → Antoine approves → Manager implements
+- New event types → Manager adds to `schemas.md` → notifies all agents via cluster message
+
+---
+
+## 6. Rollout Phases
+
+| Phase | When | What | Gate |
+|-------|------|------|------|
+| **R0: Schema** | Week 1 | Create `data-contracts/schemas.md`, `reports/templates/`, directory structure | Manager reviews, Antoine approves structure |
+| **R1: Logging** | Week 1-2 | All active agents adopt structured log format in `project_log.md` | 48h of clean structured logs from all agents |
+| **R2: Exec Dashboard** | Week 2 | Manager auto-generates `exec-summary.md` from logs | Antoine confirms it's useful and accurate |
+| **R3: Tech + Ops Dashboards** | Week 3 | Technical and operations dashboards go live | Tech Lead validates technical dashboard accuracy |
+| **R4: Weekly Reports** | Week 3-4 | Automated weekly report generation | First 2 weekly reports reviewed by Antoine |
+| **R5: KPI Tracking** | Week 4 | Rolling metrics in `kpi/metrics.md` | KPIs match reality for 2 consecutive weeks |
+| **R6: PDF Reports** | Week 5+ | Client-facing report generation via atomaste-reports | First PDF passes Auditor review |
+
+**Each phase has a go/no-go gate. No skipping.**
+
+---
+
+## 7. Risks & Mitigations
+
+| # | Risk | Impact | Likelihood | Mitigation |
+|---|------|--------|------------|------------|
+| 1 | **Log format drift** — agents write inconsistent entries | Dashboards break | Medium | Auditor spot-checks weekly; Manager rejects malformed entries |
+| 2 | **Information overload** — exec dashboard becomes too long | Antoine stops reading it | Medium | Hard cap: 30 lines. Ruthless prioritization. |
+| 3 | **Stale data** — dashboards not updated after agent activity | False confidence | High | Manager updates dashboards on every log synthesis cycle |
+| 4 | **Token cost explosion** — dashboard generation burns budget | Budget overrun | Low | Dashboard gen is cheap (small files). Monitor via ops dashboard. |
+| 5 | **Single point of failure** — Manager agent owns all dashboards | Manager down = no visibility | Medium | Raw `project_log.md` always available; any agent can read it |
+| 6 | **Scope creep** — adding features before basics work | Delayed delivery | High | Strict phase gates. No R3 until R2 is validated. |
+| 7 | **File conflicts** — multiple agents writing simultaneously | Data corruption | Low | Only Manager writes dashboards; log is append-only with timestamps |
+
+---
+
+## 8. KPIs & Gate Rules
+
+### KPI List
+
+| # | KPI | Target | Measurement |
+|---|-----|--------|-------------|
+| K1 | Dashboard freshness | ≤1h stale | Time since last exec-summary update |
+| K2 | Log compliance rate | ≥95% | % of log entries matching schema |
+| K3 | Weekly report delivery | 100% on-time | Generated by Friday 17:00 EST |
+| K4 | CEO read-time | ≤60 seconds | Exec summary length ≤30 lines |
+| K5 | Decision backlog age | ≤48h | Max age of unresolved `decision-needed` events |
+| K6 | Project status accuracy | No surprises | Zero cases where dashboard says green but reality is red |
+| K7 | Agent error rate | ≤5% | Failed tasks / total tasks per agent per week |
+| K8 | Report generation cost | ≤$2/week | Token cost for all dashboard + report generation |
+
+### Gate Rules
+
+| Gate | Criteria | Evaluator |
+|------|----------|-----------|
+| **G1: Schema Approved** | Antoine signs off on data contracts + directory structure | Antoine |
+| **G2: Logging Stable** | 48h of compliant logs from all active agents, ≥95% schema compliance | Auditor |
+| **G3: Exec Dashboard Valid** | Antoine confirms dashboard matches his understanding of project state | Antoine |
+| **G4: Full Dashboards Live** | All 3 dashboards updating correctly for 1 week | Manager + Tech Lead |
+| **G5: Reports Automated** | 2 consecutive weekly reports generated without manual intervention | Manager |
+| **G6: System Mature** | All KPIs met for 2 consecutive weeks | Antoine (final sign-off) |
+
+---
+
+## Decision Required
+
+**Antoine:** Approve this plan to begin R0 (schema creation) immediately, or flag sections needing revision.
+
+**Estimated total effort:** ~15 agent-hours across 5 weeks. Zero new infrastructure. Zero new dependencies.