### Frontend Setup and Development Server

Source: https://github.com/cherryaugusta/consumer-duty-evidence-engine/blob/main/README.md

Commands to install frontend dependencies using pnpm and start the Vite development server.

```powershell
cd frontend
pnpm install
pnpm dev
```

--------------------------------

### Backend Setup Commands

Source: https://github.com/cherryaugusta/consumer-duty-evidence-engine/blob/main/README.md

Commands to set up the Python backend environment, including activating a virtual environment and installing development dependencies.

```powershell
cd D:\AI-Projects\consumer-duty-evidence-engine
.\.venv\Scripts\Activate.ps1
python -m pip install -r backend\requirements\dev.txt
```

--------------------------------

### Start Django Backend Server

Source: https://github.com/cherryaugusta/consumer-duty-evidence-engine/blob/main/docs/architecture/observability.md

Command to navigate to the backend directory and start the Django development server.

```powershell
cd backend
python manage.py runserver
```

--------------------------------

### Start Docker Compose Services

Source: https://github.com/cherryaugusta/consumer-duty-evidence-engine/blob/main/docs/adr/0001-monolith-with-async-workers.md

Command to start the database and Redis services using Docker Compose. Redis is used as the message broker and result backend.

```bash
docker compose up -d db redis
```

--------------------------------

### Run Backend Migrations and Server

Source: https://github.com/cherryaugusta/consumer-duty-evidence-engine/blob/main/README.md

Commands to apply database migrations and start the Django development server.

```powershell
cd backend
python manage.py migrate
python manage.py runserver
```

--------------------------------

### Idempotency Check Example

Source: https://github.com/cherryaugusta/consumer-duty-evidence-engine/blob/main/docs/architecture/ingestion-pipeline.md

An example demonstrating how to ensure task idempotency by checking if the output already exists before proceeding with processing.

```python
if case.parsed_output_exists:
    return
```

--------------------------------

### Django Logging Configuration

Source: https://github.com/cherryaugusta/consumer-duty-evidence-engine/blob/main/docs/architecture/observability.md

Example of configuring JSON formatted logging in Django settings.

```python
LOGGING = {
    "version": 1,
    "disable_existing_loggers": False,
    "formatters": {
        "json": {
            "format": "% (message)s"
        }
    },
    "handlers": {
        "console": {
            "class": "logging.StreamHandler",
            "formatter": "json"
        }
    },
    "root": {
        "handlers": ["console"],
        "level": "INFO"
    }
}
```

--------------------------------

### Example Structured Log Entry

Source: https://github.com/cherryaugusta/consumer-duty-evidence-engine/blob/main/docs/architecture/observability.md

Shows a sample structured log entry with all required fields for observability.

```json
{
  "timestamp": "2026-03-28T08:00:00Z",
  "level": "INFO",
  "logger": "apps.parsing",
  "message": "Parsing started",
  "case_id": "uuid",
  "correlation_id": "uuid",
  "stage": "parsing"
}
```

--------------------------------

### Degraded Mode Log Example

Source: https://github.com/cherryaugusta/consumer-duty-evidence-engine/blob/main/docs/architecture/observability.md

Example of a structured log event indicating that the system is operating in degraded mode due to external dependency issues.

```json
{
  "event_type": "case.assessment.completed",
  "metadata": {
    "degraded_mode": true
  }
}
```

--------------------------------

### Case Extraction Failure Log Example

Source: https://github.com/cherryaugusta/consumer-duty-evidence-engine/blob/main/docs/architecture/observability.md

Example of a structured log event when a case extraction fails, including the error type and relevant IDs.

```json
{
  "event_type": "case.extraction.failed",
  "case_id": "uuid",
  "correlation_id": "uuid",
  "metadata": {
    "error": "schema validation failed"
  }
}
```

--------------------------------

### WebSocket Event Payload Example

Source: https://github.com/cherryaugusta/consumer-duty-evidence-engine/blob/main/docs/architecture/state-machine.md

Example JSON payload for a WebSocket event triggered by a state transition. This payload includes case identifiers, status information, and timestamps for real-time UI updates and observability.

```json
{
  "case_id": "<uuid>",
  "status": "assessment_pending",
  "review_status": "unassigned",
  "timestamp": "<iso8601>",
  "degraded_mode_active": false
}
```

--------------------------------

### Structured LLM Output Example

Source: https://github.com/cherryaugusta/consumer-duty-evidence-engine/blob/main/docs/adr/0004-structured-output-first.md

Demonstrates a structured JSON output from an AI component, including a claim ID, issue type, confidence score, and a structured explanation. This format is machine-readable and schema-valid.

```json
{
  "claim_id": "uuid",
  "issue_type": "misleading_information",
  "confidence": 0.87,
  "explanation": "Fees are mentioned but not clearly disclosed upfront"
}
```

--------------------------------

### Unstructured LLM Output Example

Source: https://github.com/cherryaugusta/consumer-duty-evidence-engine/blob/main/docs/adr/0004-structured-output-first.md

Illustrates a typical free-form text output from a language model, which lacks structure and is difficult to process programmatically.

```text
"The claim appears misleading because fees are not clearly disclosed."
```

--------------------------------

### PostgreSQL Environment Variables for Local Development

Source: https://github.com/cherryaugusta/consumer-duty-evidence-engine/blob/main/docs/adr/0002-postgresql-primary-store.md

Sets essential environment variables for connecting to a local PostgreSQL database instance for the project. Ensure these match your local setup.

```powershell
$env:POSTGRES_HOST="localhost"
$env:POSTGRES_PORT="55432"
$env:POSTGRES_DB="cdee"
$env:POSTGRES_USER="cdee"
$env:POSTGRES_PASSWORD="cdee"
$env:DJANGO_SETTINGS_MODULE="config.settings.local"
```

--------------------------------

### Example Correlation ID

Source: https://github.com/cherryaugusta/consumer-duty-evidence-engine/blob/main/docs/architecture/observability.md

Illustrates the structure of a correlation ID object used for tracing cases.

```json
{
  "correlation_id": "case-1234-uuid",
  "case_id": "uuid",
  "stage": "parsing"
}
```

--------------------------------

### Run Celery Worker

Source: https://github.com/cherryaugusta/consumer-duty-evidence-engine/blob/main/README.md

Command to start a Celery worker for asynchronous task processing.

```powershell
cd backend
celery -A config worker -l info
```

--------------------------------

### Celery Worker Command

Source: https://github.com/cherryaugusta/consumer-duty-evidence-engine/blob/main/docs/architecture/ingestion-pipeline.md

Command to start a Celery worker in Windows-compatible mode. Requires specifying the Celery application and logging level.

```bash
celery -A config worker -l info -P solo
```

--------------------------------

### Run Celery Worker

Source: https://github.com/cherryaugusta/consumer-duty-evidence-engine/blob/main/docs/adr/0001-monolith-with-async-workers.md

Command to start a Celery worker for the application. The `-P solo` pool is used to ensure compatibility with Windows.

```powershell
python -m celery -A config worker -l info -P solo
```

--------------------------------

### WebSocket Update Payload

Source: https://github.com/cherryaugusta/consumer-duty-evidence-engine/blob/main/docs/architecture/ingestion-pipeline.md

Example JSON payload for WebSocket updates, providing real-time status information about a case transition. Includes case ID, status, and timestamp.

```json
{
  "case_id": "<uuid>",
  "status": "parsing",
  "timestamp": "<iso8601>"
}
```

--------------------------------

### Seed Demo Data

Source: https://github.com/cherryaugusta/consumer-duty-evidence-engine/blob/main/README.md

Run this command to populate the system with 12 seeded demo cases for various scenarios.

```powershell
python infra/scripts/seed_demo_data.py
```

--------------------------------

### Run Evaluation Suite

Source: https://github.com/cherryaugusta/consumer-duty-evidence-engine/blob/main/README.md

Execute the synthetic evaluation harness with over 40 cases to assess system performance across different scenarios.

```powershell
python infra/scripts/run_eval_suite.py
```

--------------------------------

### Metrics Overview Endpoint

Source: https://github.com/cherryaugusta/consumer-duty-evidence-engine/blob/main/README.md

Provides an overview of system metrics.

```APIDOC
## GET /api/metrics/overview/

### Description
Retrieves an overview of system metrics.

### Method
GET

### Endpoint
/api/metrics/overview/
```

--------------------------------

### High-Level System Architecture Diagram

Source: https://github.com/cherryaugusta/consumer-duty-evidence-engine/blob/main/docs/architecture/system-overview.md

Illustrates the interaction between the React UI, Django API, PostgreSQL, Redis, Celery Worker, WebSockets, Audit Events, and Eval Runner.

```mermaid
flowchart LR
    UI[React UI] --> API[Django DRF API]
    API --> DB[(PostgreSQL)]
    API --> REDIS[(Redis)]
    API --> WS[Channels / WebSockets]
    API --> CELERY[Celery Worker]

    CELERY --> DB
    CELERY --> REDIS

    API --> AUDIT[Audit Events]
    API --> EVALS[Eval Runner]

```

--------------------------------

### Task Orchestration Pattern

Source: https://github.com/cherryaugusta/consumer-duty-evidence-engine/blob/main/docs/architecture/ingestion-pipeline.md

Demonstrates the pattern of explicit task chaining via enqueueing for pipeline stages. This allows for simpler debugging and better retry control.

```python
def parse_artifact_task(case_id):
    # process
    extract_case_task.delay(case_id)
```

--------------------------------

### Swagger UI Endpoint

Source: https://github.com/cherryaugusta/consumer-duty-evidence-engine/blob/main/README.md

Provides access to the Swagger UI for interactive API documentation.

```APIDOC
## GET /api/docs/

### Description
Provides access to the Swagger UI, an interactive documentation interface for the API.

### Method
GET

### Endpoint
/api/docs/
```

--------------------------------

### Failure Handling in Tasks

Source: https://github.com/cherryaugusta/consumer-duty-evidence-engine/blob/main/docs/architecture/ingestion-pipeline.md

Illustrates the try/except block used for failure handling within pipeline tasks. It ensures the case status is updated to FAILED upon an exception.

```python
try:
    ...
except Exception:
    case.status = FAILED
    case.save()
```

--------------------------------

### Enqueue Parsing Task

Source: https://github.com/cherryaugusta/consumer-duty-evidence-engine/blob/main/docs/architecture/ingestion-pipeline.md

Initiates the artifact parsing process by enqueuing a Celery task. This is the entry point for processing uploaded artifacts.

```python
parse_artifact_task.delay(case_id)
```

--------------------------------

### Review Status Lifecycle Diagram

Source: https://github.com/cherryaugusta/consumer-duty-evidence-engine/blob/main/docs/architecture/state-machine.md

Visual representation of the allowed transitions for review tasks, independent of the case status. This diagram outlines the workflow from unassigned tasks to their final closed state.

```mermaid
stateDiagram-v2
    [*] --> UNASSIGNED

    UNASSIGNED --> ASSIGNED
    ASSIGNED --> IN_REVIEW

    IN_REVIEW --> APPROVED
    IN_REVIEW --> OVERRIDDEN
    IN_REVIEW --> ESCALATED

    APPROVED --> CLOSED
    OVERRIDDEN --> CLOSED
    ESCALATED --> CLOSED

```

--------------------------------

### Recommendation Endpoint

Source: https://github.com/cherryaugusta/consumer-duty-evidence-engine/blob/main/README.md

Provides access to the recommendation for a specific case.

```APIDOC
## GET /api/cases/{id}/recommendation/

### Description
Retrieves the recommendation for a specific case identified by its ID.

### Method
GET

### Endpoint
/api/cases/{id}/recommendation/

### Parameters
#### Path Parameters
- **id** (string) - Required - The unique identifier of the case.
```

--------------------------------

### Enable pgvector Extension in PostgreSQL

Source: https://github.com/cherryaugusta/consumer-duty-evidence-engine/blob/main/docs/adr/0003-pgvector-limited-retrieval.md

This command enables the PostgreSQL 'vector' extension, which is necessary for using pgvector. It's typically run during database initialization or for test environments.

```powershell
docker compose exec db psql -U cdee -d template1 -c "CREATE EXTENSION IF NOT EXISTS vector;"
```

--------------------------------

### Review Tasks Endpoint

Source: https://github.com/cherryaugusta/consumer-duty-evidence-engine/blob/main/README.md

Provides access to review tasks.

```APIDOC
## GET /api/review-tasks/

### Description
Retrieves a list of review tasks.

### Method
GET

### Endpoint
/api/review-tasks/
```

--------------------------------

### Case Status Lifecycle Diagram

Source: https://github.com/cherryaugusta/consumer-duty-evidence-engine/blob/main/docs/architecture/state-machine.md

Visual representation of the allowed transitions between different case statuses in the evidence review workflow. This diagram helps understand the progression from initial states to final states like ARCHIVED.

```mermaid
stateDiagram-v2
    [*] --> NEW

    NEW --> INGESTION_PENDING
    NEW --> FAILED

    INGESTION_PENDING --> PARSING
    INGESTION_PENDING --> FAILED

    PARSING --> PARSED
    PARSING --> FAILED

    PARSED --> EXTRACTION_PENDING
    PARSED --> FAILED

    EXTRACTION_PENDING --> EXTRACTED
    EXTRACTION_PENDING --> NEEDS_REVIEW
    EXTRACTION_PENDING --> FAILED

    EXTRACTED --> MAPPING_PENDING
    EXTRACTED --> FAILED

    MAPPING_PENDING --> MAPPED
    MAPPING_PENDING --> FAILED

    MAPPED --> ASSESSMENT_PENDING
    MAPPED --> FAILED

    ASSESSMENT_PENDING --> ASSESSED
    ASSESSMENT_PENDING --> NEEDS_REVIEW
    ASSESSMENT_PENDING --> FAILED

    ASSESSED --> APPROVED
    ASSESSED --> NEEDS_REVIEW
    ASSESSED --> ESCALATED

    NEEDS_REVIEW --> APPROVED
    NEEDS_REVIEW --> ESCALATED
    NEEDS_REVIEW --> ARCHIVED

    APPROVED --> ARCHIVED
    ESCALATED --> ARCHIVED

    FAILED --> INGESTION_PENDING
    FAILED --> PARSING

    ARCHIVED --> [*]

```

--------------------------------

### Logging Celery Tasks with Case IDs

Source: https://github.com/cherryaugusta/consumer-duty-evidence-engine/blob/main/docs/architecture/observability.md

Use this pattern to include `case_id` and `correlation_id` in every task log for better traceability. Ensure the logger object is properly configured.

```python
logger.info(
    "Extraction completed",
    extra={
        "case_id": case.id,
        "correlation_id": case.correlation_id,
        "stage": "extraction"
    }
)
```

--------------------------------

### Cases Endpoint

Source: https://github.com/cherryaugusta/consumer-duty-evidence-engine/blob/main/README.md

Provides access to case-related information.

```APIDOC
## GET /api/cases/

### Description
Retrieves a list of cases.

### Method
GET

### Endpoint
/api/cases/
```

--------------------------------

### API Schema Endpoint

Source: https://github.com/cherryaugusta/consumer-duty-evidence-engine/blob/main/README.md

Provides access to the OpenAPI schema for the API.

```APIDOC
## GET /api/schema/

### Description
Retrieves the OpenAPI schema definition for the API.

### Method
GET

### Endpoint
/api/schema/
```

--------------------------------

### Evals Runs Endpoint

Source: https://github.com/cherryaugusta/consumer-duty-evidence-engine/blob/main/README.md

Provides access to evaluation runs.

```APIDOC
## GET /api/evals/runs/

### Description
Retrieves a list of evaluation runs.

### Method
GET

### Endpoint
/api/evals/runs/
```

--------------------------------

### Django Model for Audit Events

Source: https://github.com/cherryaugusta/consumer-duty-evidence-engine/blob/main/docs/architecture/observability.md

Defines the database model for storing audit events. It includes fields for event type, case association, correlation ID, timestamp, and flexible metadata.

```python
class AuditEvent(models.Model):
    event_type = models.CharField(max_length=255)
    case = models.ForeignKey("Case", on_delete=models.CASCADE)
    correlation_id = models.UUIDField()
    timestamp = models.DateTimeField(auto_now_add=True)
    metadata = models.JSONField(default=dict)
```

--------------------------------

### Claims Endpoint

Source: https://github.com/cherryaugusta/consumer-duty-evidence-engine/blob/main/README.md

Provides access to claims associated with a specific case.

```APIDOC
## GET /api/cases/{id}/claims/

### Description
Retrieves a list of claims for a specific case identified by its ID.

### Method
GET

### Endpoint
/api/cases/{id}/claims/

### Parameters
#### Path Parameters
- **id** (string) - Required - The unique identifier of the case.
```

--------------------------------

### Audit Event Structure

Source: https://github.com/cherryaugusta/consumer-duty-evidence-engine/blob/main/docs/architecture/observability.md

This JSON structure represents an audit event, providing a permanent record of system actions. It includes event type, case identifiers, timestamp, and metadata.

```json
{
  "event_type": "case.parsing.completed",
  "case_id": "uuid",
  "correlation_id": "uuid",
  "timestamp": "2026-03-28T08:01:00Z",
  "metadata": {
    "duration_ms": 1200
  }
}
```

--------------------------------

### Assessments Endpoint

Source: https://github.com/cherryaugusta/consumer-duty-evidence-engine/blob/main/README.md

Provides access to assessments associated with a specific case.

```APIDOC
## GET /api/cases/{id}/assessments/

### Description
Retrieves a list of assessments for a specific case identified by its ID.

### Method
GET

### Endpoint
/api/cases/{id}/assessments/

### Parameters
#### Path Parameters
- **id** (string) - Required - The unique identifier of the case.
```

=== COMPLETE CONTENT === This response contains all available snippets from this library. No additional content exists. Do not make further requests.