## Project Ploughshares — OSINT Intake Workflow
OSINT intake for international trade signals to help detect embargo‑evasion.

---

### Problem
- Fragmented intake
- Missed signals at scale
- Limited auditability

---

### Approach
- Unified intake + approval (Flask UI + JSON API)
- Automated ingestion (crawlers, curated feeds)
- Quality & accessibility checks in dev

---

### Architecture
```mermaid
flowchart LR
    Crawler[Crawlers & Feeds] -->|POST /ingest| API
    Analyst -->|Submit| WebUI
    WebUI -->|REST| API
    API --> DB[(DB)]
    API --> Queue[Approval Queue]
    Queue -->|Approve/Deny| API
```

---

### End-to-End
```mermaid
sequenceDiagram
    participant Analyst
    participant WebUI as Web UI
    participant API
    participant DB as DB
    participant Crawler

    Analyst->>WebUI: Submit signal
    WebUI->>API: Create (pending)
    API->>DB: Persist
    Crawler->>API: Ingest leads
    API->>DB: Upsert
    Analyst->>WebUI: Approve/Deny
    WebUI->>API: Decision
    API->>DB: Finalize
```

---

### Implemented
- Flask app (`docker/ploughshares/app.py`)
- Schema/init (`docker/ploughshares/schema.sql`, `init_db.py`)
- Crawlers/feeds (`docker/crawler/*`, `docker/crawler_dorks/*`)
- Quality checks (`check_accessibility.py`, `check_contrast.py`, `run_lighthouse_test.sh`)
- Tests (`tests/`)

---

### Next
- Implement vector search for intake to perform semantic matching beyond literal keywords and synonyms, surfacing adjacent signals and concepts.
- Author and test Disaster Recovery Plan (DRP) and Business Continuity Plan (BCP), with RTO/RPO targets, backups, failover, and runbooks.
- Expand API docs and sample payloads.