refactor(arch): promote workspace/ to root level (L1 template, L2 populated)

BREAKING: workspace/ moved from knowledge/workspace/ to root level.
This aligns with the architectural decision that business data is a
first-class citizen alongside knowledge/ and agents/, not nested under it.

- workspace/ now at root: _org, _team, _finance, _meetings, _automations, _tools, inbox/
- core/paths.py: WORKSPACE = ROOT / "workspace" (replaces KNOWLEDGE_WORKSPACE)
- bucket_processor.py: updated imports to use WORKSPACE
- directory-contract.md: updated architecture diagram and directory table
- reference/SOURCE-TREE.md: updated top-level structure
- .gitignore: workspace/ tracked as L1 template, L2 populated

Layer behavior:
  L1: workspace/ structure (template dirs, index.md, setup docs)
  L2: workspace/ populated (real business data)
  L3: knowledge/personal/ remains gitignored

79 tests pass. Ruff clean.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

Author: thiagofinch

.claude/rules/directory-contract.md

@@ -7,7 +7,7 @@
 
 ---
 
-## Arquitetura Tridimensional (knowledge/)
+## Arquitetura Tridimensional (3 Buckets)
 
 ```
 knowledge/
@@ -17,20 +17,21 @@ knowledge/
 │   ├── playbooks/      → Actionable playbooks
 │   ├── sources/        → Source compilations
 │   └── inbox/          → Raw expert materials
-├── workspace/      ← Bucket 2: Business Data (L2/L3)
-│   ├── _org/           → Organization structure
-│   ├── _team/          → Team data
-│   ├── _finance/       → Financial data (L3)
-│   ├── _meetings/      → Meeting notes
-│   ├── _automations/   → Tool configs
-│   ├── _tools/         → Detected tools log
-│   └── inbox/          → Raw business materials
 └── personal/       ← Bucket 3: Cognitive/Private (L3 ONLY)
     ├── _email/         → Email digests
     ├── _messages/      → WhatsApp/Slack
     ├── _calls/         → Call transcripts
     ├── _cognitive/     → Mental models, reflections
     └── inbox/          → Raw personal materials
+
+workspace/              ← Bucket 2: Business Data (ROOT level, L1 template / L2 populated)
+├── _org/               → Organization structure
+├── _team/              → Team data
+├── _finance/           → Financial data
+├── _meetings/          → Meeting notes
+├── _automations/       → Tool configs
+├── _tools/             → Detected tools log
+└── inbox/              → Raw business materials
 ```
 
 ## Diretórios e Propósito
@@ -47,8 +48,8 @@ knowledge/
 | `artifacts/` | Generated Output | audit reports, validation | Gitignored |
 | `logs/` | Session Logs | batches, JSONL audit trails | Gitignored |
 | `inbox/` | Raw Materials | L3 personal content | Gitignored |
-| `knowledge/external/` | Bucket 1 | Expert dna, dossiers, playbooks | Gitignored |
-| `knowledge/workspace/` | Bucket 2 | Business data, org, finance | Gitignored |
+| `workspace/` | Bucket 2 | Business data, org, finance | Tracked (L1 template, L2 populated) |
+| `knowledge/external/` | Bucket 1 | Expert dna, dossiers, playbooks | Tracked (L2) |
 | `knowledge/personal/` | Bucket 3 | Cognitive, email, calls | Gitignored (L3) |
 | `research/` | Ad-hoc Analysis | L3 blueprints, deep-dives | Gitignored |
 | `processing/` | Pipeline Artifacts | speakers, entities, diarization | Gitignored |
@@ -70,7 +71,7 @@ knowledge/
 | `nav_map_builder.py` | `knowledge/external/` | `ROUTING["nav_map"]` |
 | `sow_generator.py` | `agents/sua-empresa/sow/` | `ROUTING["sow_output"]` |
 | `organized_downloader.py` | `inbox/` | `ROUTING["download"]` |
-| *(workspace scripts)* | `knowledge/workspace/` | `ROUTING["workspace_data"]` |
+| *(workspace scripts)* | `workspace/` | `ROUTING["workspace_data"]` |
 | *(personal scripts)* | `knowledge/personal/` | `ROUTING["personal_data"]` |
 
 ## RAG Isolation
@@ -97,7 +98,7 @@ from core.paths import ROUTING, KNOWLEDGE_EXTERNAL, KNOWLEDGE_WORKSPACE, KNOWLED
 # Correto: usar constante
 output = ROUTING["audit_report"] / "report.json"
 dna_path = KNOWLEDGE_EXTERNAL / "dna" / "persons" / "alex-hormozi"
-workspace = ROUTING["workspace_data"] / "_meetings"
+workspace = ROUTING["workspace_data"] / "_meetings"  # resolves to workspace/_meetings
 
 # Errado: hardcodar path
 output = Path("knowledge/dna/persons/alex-hormozi")  # PROIBIDO (stale path)

.gitignore

@@ -448,15 +448,9 @@ squads/
 !knowledge/external/inbox/
 !knowledge/external/inbox/.gitkeep
 
-# Knowledge workspace scaffolding L1/L2
-!knowledge/workspace/
-!knowledge/workspace/index.md
-!knowledge/workspace/MISSING-CONTEXT-LOG.md
-!knowledge/workspace/DETECTED-TOOLS-LOG.md
-!knowledge/workspace/SETUP-PENDING.md
-!knowledge/workspace/MASTER-INDEX.md
-!knowledge/workspace/inbox/
-!knowledge/workspace/inbox/.gitkeep
+# Workspace — root level (L1 template, L2 populated)
+!workspace/
+!workspace/**
 
 # Knowledge personal — L3 ONLY (never commit content)
 # The personal/.gitignore blocks everything except itself and index.md
@@ -502,8 +496,8 @@ squads/
 !knowledge/external/TAG-RESOLVER.json
 !knowledge/external/inbox/**
 
-# Knowledge workspace populado
-!knowledge/workspace/**
+# Workspace populado (root level, not inside knowledge/)
+# Already covered by !workspace/** in BLOCO 4
 
 # Expert agents populados
 !agents/minds/**

core/intelligence/pipeline/bucket_processor.py

@@ -26,9 +26,9 @@
 from core.paths import (
     KNOWLEDGE_EXTERNAL,
     KNOWLEDGE_PERSONAL,
-    KNOWLEDGE_WORKSPACE,
     LOGS,
     ROUTING,
+    WORKSPACE,
 )
 
 logger = logging.getLogger(__name__)
@@ -52,18 +52,18 @@
         },
     },
     "workspace": {
-        "path": KNOWLEDGE_WORKSPACE,
+        "path": WORKSPACE,
         "inbox": ROUTING["workspace_inbox"],
         "color": "red",
         "label": "Business Intelligence",
         "layer": "L2/L3",
         "subdirs": {
-            "org": KNOWLEDGE_WORKSPACE / "_org",
-            "team": KNOWLEDGE_WORKSPACE / "_team",
-            "finance": KNOWLEDGE_WORKSPACE / "_finance",
-            "meetings": KNOWLEDGE_WORKSPACE / "_meetings",
-            "automations": KNOWLEDGE_WORKSPACE / "_automations",
-            "tools": KNOWLEDGE_WORKSPACE / "_tools",
+            "org": WORKSPACE / "_org",
+            "team": WORKSPACE / "_team",
+            "finance": WORKSPACE / "_finance",
+            "meetings": WORKSPACE / "_meetings",
+            "automations": WORKSPACE / "_automations",
+            "tools": WORKSPACE / "_tools",
         },
     },
     "personal": {

core/paths.py

@@ -24,14 +24,15 @@
 SKILLS = CLAUDE / "skills"
 HOOKS = CLAUDE / "hooks"
 COMMANDS = CLAUDE / "commands"
+WORKSPACE = ROOT / "workspace"  # Business data (L1 template, L2 populated)
 
 # ── GITIGNORED (L3 / Runtime) ────────────────────────────────────
 LOGS = ROOT / "logs"
 INBOX = ROOT / "inbox"
 KNOWLEDGE = ROOT / "knowledge"
 KNOWLEDGE_EXTERNAL = KNOWLEDGE / "external"
-KNOWLEDGE_WORKSPACE = KNOWLEDGE / "workspace"
 KNOWLEDGE_PERSONAL = KNOWLEDGE / "personal"
+# KNOWLEDGE_WORKSPACE removed — workspace is now at root level (WORKSPACE)
 PROCESSING = ROOT / "processing"
 ARTIFACTS = ROOT / "artifacts"
 DATA = ROOT / ".data"
@@ -88,11 +89,11 @@
     # Trash (never delete, always move here)
     "trash": TRASH,
     # Knowledge buckets
-    "workspace_data": KNOWLEDGE_WORKSPACE,
+    "workspace_data": WORKSPACE,
     "personal_data": KNOWLEDGE_PERSONAL,
     "rag_expert": RAG_EXPERT,
     "rag_business": RAG_BUSINESS,
-    "workspace_inbox": KNOWLEDGE_WORKSPACE / "inbox",
+    "workspace_inbox": WORKSPACE / "inbox",
     "personal_inbox": KNOWLEDGE_PERSONAL / "inbox",
     "external_inbox": KNOWLEDGE_EXTERNAL / "inbox",
     # Log templates (L1 — mechanism, not data)

reference/SOURCE-TREE.md

@@ -15,10 +15,10 @@ mega-brain/
 ├── agents/               Knowledge agents (conclave, cargo, minds)
 ├── reference/            Documentation (guides, standards, ADRs)
 ├── docs/                 Legacy docs (migrating to reference/)
-├── knowledge/            Knowledge base (3 buckets)
+├── workspace/            Business data — L1 template, L2 populated (org, finance, meetings)
+├── knowledge/            Knowledge base (2 buckets)
 │   ├── external/         Expert content (dna, dossiers, playbooks)
-│   ├── workspace/        Business data (org, finance, meetings)
-│   └── personal/         Private content (email, calls, cognitive)
+│   └── personal/         Private content — L3 only (email, calls, cognitive)
 ├── artifacts/            Pipeline output (chunks, insights, canonical)
 ├── inbox/                Raw materials awaiting processing
 ├── logs/                 Session and batch logs
@@ -41,9 +41,9 @@ mega-brain/
 
 | Layer | Content | Git | npm |
 |-------|---------|-----|-----|
-| L1 (Community) | core/, bin/, agents/templates, .claude/ | Tracked | Published |
-| L2 (Pro) | agents/cargo/*, knowledge/external/* | Tracked | Premium |
-| L3 (Personal) | inbox/*, logs/*, .env, sessions | Gitignored | Never |
+| L1 (Community) | core/, bin/, agents/templates, .claude/, workspace/ (template) | Tracked | Published |
+| L2 (Pro) | agents/cargo/*, knowledge/external/*, workspace/ (populated) | Tracked | Premium |
+| L3 (Personal) | inbox/*, logs/*, knowledge/personal/*, .env, sessions | Gitignored | Never |
 | NEVER | .env, credentials, .DS_Store, __pycache__ | Blocked | Blocked |
 
 ---

workspace/DETECTED-TOOLS-LOG.md

@@ -0,0 +1,28 @@
+# Detected Tools Log
+
+> **Purpose:** Track tools and platforms detected during sessions.
+> **Location:** `knowledge/workspace/DETECTED-TOOLS-LOG.md`
+> **Updated by:** JARVIS when tools/platforms are mentioned in conversations.
+
+## How This Works
+
+When the user mentions a tool, platform, or SaaS product that is part of their
+business stack, JARVIS logs it here. Over time this builds a complete picture of the
+business technology stack, which helps agents make tool-aware recommendations.
+
+## Format
+
+Each entry follows this structure:
+
+```
+### [Tool Name]
+- **Category:** [CRM / Automation / Finance / Communication / Analytics / Other]
+- **First detected:** YYYY-MM-DD
+- **Integration status:** [Active / Planned / Deprecated / Unknown]
+- **MCP available:** [Yes / No / N/A]
+- **Notes:** [Any context about how the tool is used]
+```
+
+## Detected Tools
+
+_No tools detected yet. Tools will be logged here as they are mentioned in sessions._

workspace/MASTER-INDEX.md

@@ -0,0 +1,49 @@
+# Workspace Master Index
+
+> **Purpose:** Human-readable index of all workspace content.
+> **Location:** `knowledge/workspace/MASTER-INDEX.md`
+> **Updated by:** JARVIS after any workspace content changes.
+
+## Directory Overview
+
+| Directory | Content Count | Last Updated |
+|-----------|--------------|--------------|
+| `_org/` | 0 items | - |
+| `_team/` | 0 items | - |
+| `_finance/` | 0 items | - |
+| `_meetings/` | 0 items | - |
+| `_tools/` | 0 items | - |
+| `_automations/` | 0 items | - |
+| `inbox/` | 0 items | - |
+
+## Content by Category
+
+### Organization (`_org/`)
+_No content yet._
+
+### Team (`_team/`)
+_No content yet._
+
+### Finance (`_finance/`)
+_No content yet._
+
+### Meetings (`_meetings/`)
+_No content yet._
+
+### Tools (`_tools/`)
+_No content yet._
+
+### Automations (`_automations/`)
+_No content yet._
+
+## Related Files
+
+- [Missing Context Log](./MISSING-CONTEXT-LOG.md) -- Gaps detected by agents
+- [Detected Tools Log](./DETECTED-TOOLS-LOG.md) -- Business tech stack
+- [Setup Pending](./SETUP-PENDING.md) -- Integration backlog
+
+## Statistics
+
+- **Total workspace items:** 0
+- **Last full index:** Never
+- **Coverage score:** 0% (no business context imported yet)

workspace/MISSING-CONTEXT-LOG.md

@@ -0,0 +1,29 @@
+# Missing Context Log
+
+> **Purpose:** Track business context gaps detected by agents during sessions.
+> **Location:** `knowledge/workspace/MISSING-CONTEXT-LOG.md`
+> **Updated by:** JARVIS and cargo agents when they detect missing business data.
+
+## How This Works
+
+When an agent needs business context to answer a question but cannot find it in
+`knowledge/workspace/`, it logs the gap here. This creates a backlog of information
+that should be imported into the workspace bucket.
+
+## Format
+
+Each entry follows this structure:
+
+```
+### YYYY-MM-DD -- [Category]
+- **Detected by:** [Agent name]
+- **Context needed:** [What was missing]
+- **Why needed:** [What question triggered the detection]
+- **Suggested source:** [Where this data might come from]
+- **Priority:** [HIGH / MEDIUM / LOW]
+- **Status:** [OPEN / RESOLVED / WONT-FIX]
+```
+
+## Log Entries
+
+_No entries yet. Gaps will be logged here as agents detect missing business context._

workspace/SETUP-PENDING.md

@@ -0,0 +1,35 @@
+# Setup Pending
+
+> **Purpose:** Track pending setup items for workspace integrations.
+> **Location:** `knowledge/workspace/SETUP-PENDING.md`
+> **Updated by:** JARVIS during setup and configuration sessions.
+
+## How This Works
+
+When a workspace integration is identified but not yet configured, it gets logged
+here. This serves as a backlog for the @devops system agent and the user to
+progressively build out the workspace bucket's data sources.
+
+## Pending Items
+
+### Data Sources Not Yet Connected
+
+| Source | Category | Blocked By | Priority |
+|--------|----------|------------|----------|
+| _None yet_ | - | - | - |
+
+### Integrations Not Yet Configured
+
+| Integration | MCP Server | Status | Notes |
+|-------------|------------|--------|-------|
+| _None yet_ | - | - | - |
+
+### Manual Imports Needed
+
+| Data | Source Location | Target Directory | Notes |
+|------|----------------|------------------|-------|
+| _None yet_ | - | - | - |
+
+## Completed Setup Items
+
+_No items completed yet._

workspace/index.md

@@ -0,0 +1,58 @@
+# Knowledge / Workspace
+
+> **Bucket 2** -- Business context and operational data.
+> **Layer:** L2 (Pro, tracked scaffolding) / L3 (populated content gitignored)
+> **paths.py constant:** `KNOWLEDGE_WORKSPACE`
+
+## Purpose
+
+This bucket contains **business-specific** knowledge: team structure, financial data,
+meeting notes, tool configurations, and automation workflows. It provides the real-world
+context that transforms generic expert knowledge into actionable recommendations.
+
+## Structure
+
+```
+knowledge/workspace/
+  _org/             -- Organization charts, role definitions, SOWs
+  _team/            -- Team profiles, performance data, hiring pipeline
+  _finance/         -- Financial data, KPIs, DRE, unit economics
+  _meetings/        -- Meeting notes, decisions, action items
+  _tools/           -- Tool configurations, CRM setup, integrations
+  _automations/     -- n8n workflows, Zapier configs, automation docs
+  inbox/            -- Triage area for new business content
+    pending/        -- Awaiting classification
+    rejected/       -- Did not pass quality gate
+```
+
+## Key Files
+
+| File | Purpose |
+|------|---------|
+| `MASTER-INDEX.md` | Human-readable index of all workspace content |
+| `MISSING-CONTEXT-LOG.md` | Tracks business context gaps detected by agents |
+| `DETECTED-TOOLS-LOG.md` | Tools and platforms detected during sessions |
+| `SETUP-PENDING.md` | Pending setup items for workspace integrations |
+
+## Routing
+
+```python
+from core.paths import KNOWLEDGE_WORKSPACE, ROUTING
+
+ROUTING["workspace_data"]   # -> knowledge/workspace/
+ROUTING["workspace_inbox"]  # -> knowledge/workspace/inbox/
+ROUTING["rag_business"]     # -> .data/rag_business/ (index built from this bucket)
+```
+
+## Integration Points
+
+- Agents consult workspace/ before making strategic recommendations (Rule #16)
+- The `context_assembler.py` pulls from workspace/ for business-aware RAG
+- MCP servers (ClickUp, Notion) can push data into workspace/inbox/
+- Financial data here feeds the CFO agent's MEMORY.md
+
+## Security
+
+Populated content in this bucket is L3 (gitignored). Only the directory scaffolding
+and template files (.gitkeep, index.md) are tracked. Never commit actual business
+data, financial figures, or team personal information.