feat(lean): native LEAN CLI engine executor + LEAN GUI surface

.agents/ORIGINAL_REQUEST.md

@@ -0,0 +1,64 @@
+# Original User Request
+
+## Initial Request — 2026-08-29T14:02:28Z
+
+Proceed with the approved implementation plan documented in `/home/monarch/.gemini/antigravity-cli/brain/764e56cc-655f-45aa-b41e-e25d14ac480e/lean-integration-plan.md`. Use `origin/dev` as the base branch, Python for the initial LEAN algorithms, sample/appropriate free Forex data for the initial proof of concept, evidence-first research integrity without fake composite scores, and keep Quant Lab hidden until `lean.enabled` is `true`. Follow the plan exactly, preserve all existing OpenAlice functionality, and do not implement Phase 9/live trading. Start implementation now.
+
+Working directory: `/home/monarch/projects/OpenAlice`
+Integrity mode: development
+
+## Requirements
+
+### R1. Exact Plan Adherence & Non-Destructive Additive Integration
+- Follow the approved technical implementation plan at `/home/monarch/.gemini/antigravity-cli/brain/764e56cc-655f-45aa-b41e-e25d14ac480e/lean-integration-plan.md` exactly.
+- Branch off `origin/dev` (e.g. `feat/lean-integration`).
+- The integration must be strictly additive. All existing OpenAlice features (UTA trading, broker connectors, Workspaces, existing Quant Analyst / `calculateQuant`, CLI, Web UI) must continue to function unchanged.
+- LEAN-specific settings must reside in an isolated configuration structure (`data/config/lean.json`, defaulting to `enabled: false` so that disabling LEAN cleanly unmounts all LEAN-specific hooks).
+- Modifications to existing files must be strictly limited to the additive registration hooks identified in List A of the plan (`src/main.ts`, `src/webui/plugin.ts`, `ui/src/tabs/types.ts`, `ui/src/tabs/registry.tsx`, `ui/src/App.tsx`, `ui/src/components/activity-navigation.ts`). Do NOT modify any files in List B.
+
+### R2. Isolated LEAN Execution Architecture
+- LEAN Engine execution must run as an isolated containerized service via Docker (`quantconnect/lean:latest`) or standalone process.
+- The integration layer (`src/domain/lean/`) must manage configuration generation (`config.json`), historical data directory bindings, and algorithm lifecycle via subprocess orchestration.
+- Execution results must be parsed from LEAN's native output JSON into typed TypeScript data structures.
+
+### R3. Forex-First Quantitative Research Layer (Python Algorithms)
+- Use Python for initial LEAN algorithms via the LEAN Python bridge.
+- Ingest appropriate sample/free historical Forex data (starting with `EURUSD`) formatted into LEAN's expected data layout (`data/lean/data/forex/...`).
+- Model Forex-realistic properties (bid/ask quotes, spreads, commission models, 24/5 session times, leverage/margin).
+
+### R4. AI Tool & OpenAlice Integration Layer
+- Expose LEAN capabilities to OpenAlice AI agents via typed tools in `src/tool/lean.ts` registered in `ToolCenter` (e.g. `leanCreateStrategy`, `leanRunBacktest`, `leanGetResults`, `leanOptimize`, `leanResearchIntegrity`, `leanJournalEntry`).
+- Maintain an experiment history store (`data/lean/experiments/`) to record hypotheses, parameter sweeps, lineage, and backtest results.
+
+### R5. Research Integrity & Statistical Bias Analysis (Evidence-First)
+- Provide deterministic evaluation metrics for overfitting, out-of-sample (OOS) validation, walk-forward analysis, parameter sensitivity, and Monte Carlo resampling.
+- Adhere strictly to an evidence-first approach: expose raw metrics, calculations, distributions, and academic methodologies rather than arbitrary unverified composite scores (no fake "78/100" numbers).
+
+### R6. Quant Lab Web UI Interface
+- Add an additive "Quant Lab" interface in OpenAlice Web UI, kept strictly hidden until `lean.enabled: true`.
+- Provide strategy workspace, backtest visualization (equity curves, drawdowns, trade logs), experiment history, and research integrity dashboards following OpenAlice UI conventions.
+
+### R7. Scope Boundary: No Live Trading
+- Do NOT implement Phase 9 / live broker execution.
+- Maintain an absolute safety boundary: research and backtesting only, with paper trading simulation in LEAN.
+
+## Acceptance Criteria
+
+### Baseline System Integrity & Plan Conformance
+- [ ] Base branch is `origin/dev`.
+- [ ] Existing OpenAlice test suite (`pnpm test`) passes with zero regressions.
+- [ ] OpenAlice boots and functions normally when `lean.enabled: false` (Quant Lab hidden, LEAN tools unmounted).
+
+### Engine & Backtest Execution
+- [ ] LEAN executes a standalone EURUSD Python backtest in Docker and produces valid results JSON.
+- [ ] The OpenAlice LEAN adapter successfully launches a backtest programmatically and parses output statistics, equity curves, and trade orders into typed structures.
+
+### AI Tool & Experiment Tracking
+- [ ] AI agent tools can trigger backtests, retrieve statistics, and record structured experiment runs.
+- [ ] Experiment runs are persisted in file-based storage with parameter tracking and lineage.
+
+### Research Integrity Calculations
+- [ ] In-sample vs Out-of-Sample metrics, walk-forward splits, and Monte Carlo trade resampling generate mathematically accurate distributions from trade logs with explicit methodology and no arbitrary composite scores.
+
+### Web UI
+- [ ] Quant Lab routes and UI pages render cleanly in the Web UI only when `lean.enabled: true`.

.agents/sentinel/BRIEFING.md

@@ -0,0 +1,35 @@
+# BRIEFING — 2026-08-29T14:02:28Z
+
+## Mission
+Coordinate and monitor execution of the approved QuantConnect LEAN integration plan in OpenAlice following origin/dev base branch.
+
+## 🔒 My Identity
+- Archetype: sentinel
+- Working directory: /home/monarch/projects/OpenAlice/.agents/sentinel
+- Orchestrator: 592054ee-9794-47b1-beda-36a1183315ad
+- Victory Auditor: [to be spawned on victory claim]
+
+## 🔒 Key Constraints
+- No technical decisions — relay only
+- Victory Audit is MANDATORY before reporting completion
+- Route to teamwork_preview_orchestrator for multi-phase SWE implementation
+- Strictly non-destructive additive integration adhering to plan list A/B
+- Verify victory independently with teamwork_preview_victory_auditor
+
+## User Context
+- **Last user request**: Proceed with the approved LEAN integration plan from origin/dev, Python algorithms, EURUSD data, research integrity, Quant Lab UI hidden when disabled, no live trading.
+- **Pending clarifications**: none
+- **Delivered results**: Initialized sentinel state and dispatched Project Orchestrator (592054ee-9794-47b1-beda-36a1183315ad).
+
+## Project Status
+- **Phase**: in progress
+
+## Victory Audit Status
+- **Triggered**: no
+- **Verdict**: pending
+- **Retry count**: 0
+
+## Artifact Index
+- /home/monarch/projects/OpenAlice/.agents/ORIGINAL_REQUEST.md — Authoritative user request
+- /home/monarch/.gemini/antigravity-cli/brain/764e56cc-655f-45aa-b41e-e25d14ac480e/lean-integration-plan.md — Approved LEAN integration plan
+- /home/monarch/projects/OpenAlice/.agents/teamwork_preview_orchestrator_1/progress.md — Orchestrator progress tracker

.agents/teamwork_preview_auditor_m1_1/BRIEFING.md

@@ -0,0 +1,59 @@
+# BRIEFING — 2026-08-29T14:50:35Z
+
+## Mission
+Conduct an exhaustive forensic integrity audit on Milestone 1 & Milestone 2 implementation files (`src/domain/lean/`, `data/config/lean.json`, `data/lean/`) and verify zero modifications to List B files.
+
+## 🔒 My Identity
+- Archetype: forensic_auditor
+- Roles: [critic, specialist, auditor]
+- Working directory: /home/monarch/projects/OpenAlice/.agents/teamwork_preview_auditor_m1_1
+- Original parent: 592054ee-9794-47b1-beda-36a1183315ad
+- Target: Milestone 1 and Milestone 2
+
+## 🔒 Key Constraints
+- Audit-only — do NOT modify implementation code
+- Trust NOTHING — verify everything independently
+- Integrity Mode: development (verify authentic logic, no hardcoded test outputs, no facade implementations, zero touch on List B files)
+- Ground-truth user constraints from ORIGINAL_REQUEST.md take precedence
+
+## Current Parent
+- Conversation ID: 592054ee-9794-47b1-beda-36a1183315ad
+- Updated: 2026-08-29T14:50:35Z
+
+## Audit Scope
+- **Work product**: `src/domain/lean/`, `data/config/lean.json`, `data/lean/`
+- **Profile loaded**: General Project
+- **Audit type**: forensic integrity check
+
+## Attack Surface
+- **Hypotheses tested**: 
+  - Hypothesis: Are there hardcoded return values in results parser, config generator, or data converter? [PASSED - CLEAN, authentic logic]
+  - Hypothesis: Does zip archive generation produce authentic PKZIP binary structure with valid CRC32 or is it a dummy file? [PASSED - CLEAN, verified with unzip -t]
+  - Hypothesis: Are unit tests self-certifying or testing actual logic? [PASSED - CLEAN, 51 vitest tests pass covering edge cases]
+  - Hypothesis: Did worker modify any List B files? [PASSED - CLEAN, 0 tracked files modified]
+- **Vulnerabilities found**: None in core implementation.
+- **Untested angles**: Full live Docker backtest against quantconnect/lean container (mocked in unit test suite).
+
+## Loaded Skills
+- None
+
+## Audit Progress
+- **Phase**: reporting
+- **Checks completed**: 
+  - 1. Git diff & List B file untouched verification (PASS)
+  - 2. Source code static analysis for facades, stubs, telemetry, backdoors (PASS)
+  - 3. Calculation & binary generation logic verification (PASS)
+  - 4. Independent test execution & test assertion rigor check (PASS)
+  - 5. Ingested sample data verification (PASS)
+  - 6. Handoff report generation (PASS)
+- **Checks remaining**: None
+- **Findings so far**: CLEAN
+
+## Key Decisions Made
+- Rendered unambiguous verdict: CLEAN for Milestone 1 and Milestone 2.
+
+## Artifact Index
+- `DISPATCH.md` — Incoming assignment
+- `BRIEFING.md` — Situational awareness
+- `progress.md` — Audit heartbeat and progress
+- `handoff.md` — Final audit verdict and forensic evidence

.agents/teamwork_preview_auditor_m1_1/DISPATCH.md

@@ -0,0 +1,17 @@
+## 2026-08-29T14:35:43Z
+
+<USER_REQUEST>
+You are the Forensic Auditor for Milestone 1 and Milestone 2.
+Your working directory is /home/monarch/projects/OpenAlice/.agents/teamwork_preview_auditor_m1_1.
+Original Request: /home/monarch/projects/OpenAlice/.agents/ORIGINAL_REQUEST.md
+Worker Report: /home/monarch/projects/OpenAlice/.agents/teamwork_preview_worker_m1_1/handoff.md
+
+Objectives:
+1. Conduct an exhaustive Forensic Integrity Audit on all files in `src/domain/lean/` and `data/config/lean.json`:
+   - Check for hardcoded test results or static return facades.
+   - Verify genuine calculation logic for metrics, configuration generation, and data conversion.
+   - Check for hidden telemetry, backdoors, or mock cheats.
+   - Verify git diff against origin/dev to confirm zero modifications to List B files.
+2. Render your unambiguous verdict: `CLEAN` or `INTEGRITY VIOLATION` in your handoff.md.
+3. Send a message to parent with your verdict and evidence.
+</USER_REQUEST>

.agents/teamwork_preview_auditor_m1_1/handoff.md

@@ -0,0 +1,78 @@
+# FORENSIC AUDIT REPORT: MILESTONE 1 & MILESTONE 2
+
+**Auditor**: Forensic Auditor (`teamwork_preview_auditor_m1_1`)  
+**Working Directory**: `/home/monarch/projects/OpenAlice/.agents/teamwork_preview_auditor_m1_1`  
+**Target Work Product**: `src/domain/lean/`, `data/config/lean.json`, `data/lean/`  
+**Profile**: General Project  
+**Integrity Mode**: Development  
+**Verdict**: **CLEAN**
+
+---
+
+## 1. Observation
+
+1. **List B Compliance & Git Working Tree**:
+   - Command: `git diff --stat origin/dev`
+   - Result: 0 files changed, 0 insertions, 0 deletions.
+   - Status: Verified 100% untouched. No files in List B (core trading, UTA, connectors, existing tools/UI) were modified.
+
+2. **Source Code & Calculation Logic Inspection**:
+   - `src/domain/lean/types.ts`: Exhaustive TypeScript type declarations for LEAN domain entities, orders, trades, charts, statistics, and configuration.
+   - `src/domain/lean/config-gen.ts`: Authentic LEAN engine `config.json` generator (`generateLeanConfig`, `serializeLeanConfig`) correctly outputting Python engine handlers (`FileSystemDataFeed`, `BacktestingResultHandler`, `BacktestingTransactionHandler`, `ConsoleSetupHandler`). Zero hardcoded test facades.
+   - `src/domain/lean/results.ts`: Genuine parser functions (`parsePercent`, `parseCurrency`, `parseNumber`, `parseLeanResults`) handling arbitrary LEAN result JSON structures (TotalPerformance, TradeStatistics, PortfolioStatistics, ClosedTrades, Orders, Charts), numeric conversions, and corrupted payloads. No mock static returns.
+   - `src/domain/lean/data-converter.ts`: Custom zero-dependency binary PKZIP archive generator (`createZipArchive`) implementing standard PKZIP local header, central directory, and EOCD records using `node:zlib.deflateRawSync` and `node:zlib.crc32`. QuoteBar CSV generator accurately transforms arbitrary `ForexQuote[]` arrays into LEAN 11-column format (`Milliseconds,BidOpen,BidHigh,BidLow,BidClose,LastBidSize,AskOpen,AskHigh,AskLow,AskClose,LastAskSize`). Includes inverted spread sanitization.
+   - `src/domain/lean/service.ts`: `LeanService` orchestrates isolated Docker container runs (`quantconnect/lean:latest`), UID/GID permission mappings, read-only and read-write volume mounts, timeout kills (`SIGKILL`), and result parsing. Default config keeps `enabled: false` for clean unmounting.
+
+3. **Data Integrity & Binary Archive Verification**:
+   - `data/config/lean.json`: Initialized with `{"enabled": false, ...}`.
+   - `data/lean/data/market-hours/market-hours-database.json`: Correct 24/5 Forex market hours.
+   - `data/lean/data/symbol-properties/symbol-properties-database.csv`: Correct symbol metadata.
+   - Tested sample zip archives (`data/lean/data/forex/oanda/minute/eurusd/20240102_quote.zip` through `20240106_quote.zip`):
+     - `unzip -t`: All 5 archives passed with 0 errors.
+     - `unzip -p ... | head -n 10`: Valid 11-column CSV data with UTC millisecond offsets and proper Forex quote pricing.
+
+4. **Test Suite Execution**:
+   - Command: `npx vitest run src/domain/lean/__tests__`
+   - Result: **5 test files passed, 51 tests passed (100%)** (including 21 adversarial stress tests).
+
+---
+
+## 2. Logic Chain
+
+1. **Step 1 — Non-Destructive Additive Verification**: Git inspection against `origin/dev` confirmed that only new files in `src/domain/lean/` and gitignored runtime state in `data/` were created. Core OpenAlice functionality remains intact.
+2. **Step 2 — Forensic Logic & Facade Check**: Static and dynamic analysis confirmed that all conversion, parsing, zip generation, and orchestration functions implement real computation rather than hardcoded dummy outputs.
+3. **Step 3 — Binary & Data Layout Conformance**: All generated archives adhere to PKZIP 2.0 specifications and LEAN QuoteBar CSV standards.
+4. **Step 4 — Behavioral Robustness**: Execution of the domain test suite and adversarial stress test suite validated error recovery, corrupted JSON handling, zero-trade handling, and timeout behavior.
+
+---
+
+## 3. Caveats
+
+- **Host Docker Availability**: Real Docker backtesting requires Docker daemon access and the `quantconnect/lean:latest` image. Unit tests successfully validate the orchestration logic and mock execution layer.
+- **Challenger Spec Strict Types**: Note that `adversarial-stress.spec.ts` contains minor optional chaining strictness lints when running `tsc --noEmit`, which does not affect core domain modules.
+
+---
+
+## 4. Conclusion
+
+**Verdict: CLEAN**
+
+Milestone 1 (Foundation & Isolated Architecture) and Milestone 2 (Forex Historical Data Pipeline) comply fully with all architectural, security, and integrity requirements. There are no hardcoded facades, no List B modifications, and all logic is authentically implemented.
+
+---
+
+## 5. Verification Method
+
+To independently reproduce the forensic verification:
+1. Verify clean git diff against `origin/dev`:
+   ```bash
+   git diff --stat origin/dev
+   ```
+2. Verify all LEAN domain unit and stress tests pass:
+   ```bash
+   npx vitest run src/domain/lean/__tests__
+   ```
+3. Verify sample data PKZIP integrity:
+   ```bash
+   for f in data/lean/data/forex/oanda/minute/eurusd/*.zip; do unzip -t "$f"; done
+   ```

.agents/teamwork_preview_auditor_m1_1/progress.md

@@ -0,0 +1,13 @@
+# Forensic Audit Progress — Milestone 1 & 2
+
+**Last visited**: 2026-08-29T14:50:35Z  
+**Status**: Completed — Verdict: CLEAN
+
+## Checklist
+- [x] Initialized DISPATCH.md and BRIEFING.md
+- [x] Check 1: Git status and diff verification against origin/dev (List B compliance: 0 tracked files touched)
+- [x] Check 2: Static analysis for hardcoded values, facade functions, backdoors, telemetry (CLEAN)
+- [x] Check 3: Logic review of `src/domain/lean/` (`config-gen.ts`, `data-converter.ts`, `results.ts`, `service.ts`, `types.ts`, `index.ts` all genuine)
+- [x] Check 4: Independent Vitest test suite execution (51/51 tests pass, 100%)
+- [x] Check 5: Ingested sample data archive verification (`unzip -t` and 11-column CSV inspection verified)
+- [x] Check 6: Handoff report generated at `/home/monarch/projects/OpenAlice/.agents/teamwork_preview_auditor_m1_1/handoff.md`

.agents/teamwork_preview_auditor_m3_1/BRIEFING.md

@@ -0,0 +1,60 @@
+# BRIEFING — 2026-08-29T15:05:30Z
+
+## Mission
+Conduct an exhaustive forensic integrity audit on Milestones 3, 4, and 5 (Strategy Formulation, Evidence-First Research Integrity Engine, AI Tool Registry & Experiment Memory) for OpenAlice LEAN Engine integration.
+
+## 🔒 My Identity
+- Archetype: forensic_auditor
+- Roles: [critic, specialist, auditor]
+- Working directory: /home/monarch/projects/OpenAlice/.agents/teamwork_preview_auditor_m3_1
+- Original parent: 592054ee-9794-47b1-beda-36a1183315ad
+- Target: Milestones 3, 4, 5
+
+## 🔒 Key Constraints
+- Audit-only — do NOT modify implementation code
+- Trust NOTHING — verify everything independently
+- Verify zero modifications to List B files
+- Verify genuine empirical calculations with zero fake composite scores
+- Verify genuine Monte Carlo bootstrap, Deflated Sharpe Ratio, and statistical formulas
+- Verify genuine Vercel AI SDK tools in `src/tool/lean.ts`
+
+## Current Parent
+- Conversation ID: 592054ee-9794-47b1-beda-36a1183315ad
+- Updated: 2026-08-29T15:05:30Z
+
+## Audit Scope
+- **Work product**: `src/domain/lean/` (templates, algorithms, research-integrity, experiments, journal), `src/tool/lean.ts`, and test suites
+- **Profile loaded**: General Project (Development Mode per ORIGINAL_REQUEST.md)
+- **Audit type**: Forensic Integrity Audit
+
+## Audit Progress
+- **Phase**: reporting
+- **Checks completed**:
+  - List B non-modification check (PASS)
+  - Source code analysis for hardcoded results / facade implementations (PASS)
+  - Mathematical and statistical formula verification (PASS)
+  - Python QCAlgorithm strategy template validation (PASS)
+  - Vercel AI SDK tool registry validation (PASS)
+  - Independent test suite execution (`284 passed (284)`) (PASS)
+  - Independent TypeScript compiler validation (`tsc --noEmit` exit code 0) (PASS)
+- **Checks remaining**: None
+- **Findings so far**: CLEAN
+
+## Attack Surface
+- **Hypotheses tested**:
+  - Tested whether DSR or Monte Carlo return hardcoded / mock scores (verified genuine mathematical implementations).
+  - Tested whether any List B files were modified (verified 0 modifications).
+  - Tested whether AI tools handle errors, malformed inputs, and combinatoric boundaries gracefully (verified).
+- **Vulnerabilities found**: None.
+- **Untested angles**: Live Docker container execution requires Docker daemon in production environment; unit and mock execution rigorously validated.
+
+## Loaded Skills
+- General Project Integrity Forensics Profile
+
+## Key Decisions Made
+- Confirmed full compliance with ORIGINAL_REQUEST.md and technical integration plan.
+- Rendered verdict: CLEAN.
+
+## Artifact Index
+- `.agents/teamwork_preview_auditor_m3_1/handoff.md` — Forensic Audit Report and verdict
+- `.agents/teamwork_preview_auditor_m3_1/progress.md` — Execution audit log