| --- |
| title: LegalSync — AI-Platform v4 |
| emoji: ⚖️ |
| colorFrom: indigo |
| colorTo: green |
| sdk: docker |
| pinned: false |
| app_port: 7860 |
| --- |
| |
| # LegalSync · AI-Platform v4 — Workload |
|
|
| **DataRobot + Groq + Chrome Secondary Brain.** A governed, any-HTTP-container |
| Workload (DataRobot Workload API) exposing the LegalSync reference app. |
|
|
| Live UI: landing page (`/`), interactive API docs (`/docs`), health (`/health`). |
|
|
| ## What runs here |
|
|
| - **FastAPI workload** (`workload_app.py`) on port **7860**: |
| - `GET /health` — readiness + canary smoke test |
| - `POST /api/payments` — OmniPay (stripe / paypal / ach) |
| - `POST /api/transcription` — whisper pipeline (DPPA redaction) |
| - `POST /api/bond/recommend` — human-in-the-loop |
| - `POST /api/discovery/assemble` — Art 39.14 packets |
| - `POST /api/llm/chat` — DataRobot LLM Gateway / Groq routing (00-core) |
| - `GET /api/mcp/tools` — agentic MCP tool listing (40-agentic-ai) |
|
|
| ## Repo layout (v4, deduped) |
|
|
| ``` |
| . |
| ├── workload_app.py # FastAPI entry (Workload API compliant) |
| ├── config/ # 8 JSONs — SINGLE SOURCE OF TRUTH |
| ├── services/ # 9 processors + orchestrator agent |
| ├── schemas/ # 4 JSON schemas |
| ├── api/ # 3 OpenAPI YAMLs |
| ├── docs/ # legal (10), compliance (4), runbooks (7), training, marketing |
| ├── scripts/ # secondary-brain agents + cross-examine gate |
| ├── ai-platform/900-registry/ # 10 YAMLs (providers, models, tools, rollouts, …) |
| ├── datarobot-workload.yaml # dr CLI rollout (canary_then_progressive) |
| ├── lighthouserc.json # Lighthouse CI budgets (secondary brain) |
| └── .github/workflows/ # ci-secondary-brain.yml |
| ``` |
|
|
| ## Deploy |
|
|
| **Hugging Face (this Space):** Docker SDK — builds `Dockerfile`, serves on 7860. |
|
|
| **Local:** |
| ```bash |
| pip install -r requirements.txt |
| uvicorn workload_app:app --host 0.0.0.0 --port 8000 |
| # or |
| docker compose up --build # legalsync + lighthouse + prometheus + grafana |
| ``` |
|
|
| **DataRobot (Workload API — any HTTP container):** |
| ```bash |
| export DATAROBOT_API_KEY=... GROQ_API_KEY=... |
| dr workload create --name legalsync --image legalsync:latest --port 8000 |
| dr workload deploy --config datarobot-workload.yaml |
| ``` |
|
|
| > Requirement per docs.datarobot.com: the Workload API works with any container |
| > that listens on HTTP — deploy agent services, model inference, RAG pipelines, |
| > MCP servers and custom apps as governed Workloads. |
|
|
| ## Quality gate (Secondary Brain) |
|
|
| Primary (build/test) **and** Secondary Brain must pass before production: |
|
|
| - Lighthouse CI — perf ≥ 0.9 · a11y 1.0 · LCP < 2500 · CLS < 0.1 (3 runs) |
| - Chrome DevTools MCP — console / network / CWV cross-examination |
| - Gate policy — `both_primary_and_secondary_must_pass` (`rollout_secondary_brain.yaml`) |
|
|
| ## Single source of truth |
|
|
| `config/*.json` is authoritative; `docs/legal/*` and schemas *reference* config, |
| never redefine it (v4 dedupe rule). See `MERGED_TOOLBOX_v4.md` for the platform. |
|
|
