gsd-trae 1.0.1 → 1.0.2

package/refs/vbenchmark/tasks/ai-integration/rag-chatbot/pdf-qa/tests/functional/qa.test.py

@@ -1,146 +0,0 @@
-import pytest
-import httpx
-from pathlib import Path
-
-BASE_URL = "http://localhost:8000"
-FIXTURES_DIR = Path(__file__).parent.parent.parent / "fixtures"
-
-
-@pytest.fixture
-async def client():
-    async with httpx.AsyncClient(base_url=BASE_URL) as client:
-        yield client
-
-
-@pytest.fixture
-async def uploaded_document(client):
-    pdf_path = FIXTURES_DIR / "handbook.pdf"
-    with open(pdf_path, "rb") as f:
-        response = await client.post(
-            "/documents",
-            files={"file": ("handbook.pdf", f, "application/pdf")}
-        )
-    assert response.status_code == 200
-    return response.json()["documentId"]
-
-
-@pytest.mark.asyncio
-async def test_health_check(client):
-    response = await client.get("/health")
-    assert response.status_code == 200
-    assert response.json()["status"] == "healthy"
-
-
-@pytest.mark.asyncio
-async def test_upload_pdf(client):
-    pdf_path = FIXTURES_DIR / "handbook.pdf"
-    with open(pdf_path, "rb") as f:
-        response = await client.post(
-            "/documents",
-            files={"file": ("handbook.pdf", f, "application/pdf")}
-        )
-
-    assert response.status_code == 200
-    data = response.json()
-    assert "documentId" in data
-    assert "pageCount" in data
-    assert data["status"] == "processed"
-
-
-@pytest.mark.asyncio
-async def test_get_document_metadata(client, uploaded_document):
-    response = await client.get(f"/documents/{uploaded_document}")
-
-    assert response.status_code == 200
-    data = response.json()
-    assert data["documentId"] == uploaded_document
-    assert "chunkCount" in data
-    assert data["chunkCount"] > 0
-
-
-@pytest.mark.asyncio
-async def test_chat_vacation_policy(client, uploaded_document):
-    response = await client.post(
-        "/chat",
-        json={
-            "documentId": uploaded_document,
-            "question": "What is the vacation policy?"
-        }
-    )
-
-    assert response.status_code == 200
-    data = response.json()
-    assert "answer" in data
-    assert "sources" in data
-    assert len(data["sources"]) > 0
-
-    # Answer should mention vacation-related content
-    answer_lower = data["answer"].lower()
-    assert any(word in answer_lower for word in ["vacation", "pto", "days", "leave"])
-
-
-@pytest.mark.asyncio
-async def test_chat_expenses(client, uploaded_document):
-    response = await client.post(
-        "/chat",
-        json={
-            "documentId": uploaded_document,
-            "question": "How do I submit expenses?"
-        }
-    )
-
-    assert response.status_code == 200
-    data = response.json()
-    assert "answer" in data
-    assert "sources" in data
-
-    # Answer should be grounded in document
-    answer_lower = data["answer"].lower()
-    assert any(word in answer_lower for word in ["expense", "submit", "receipt", "reimbursement"])
-
-
-@pytest.mark.asyncio
-async def test_chat_includes_sources(client, uploaded_document):
-    response = await client.post(
-        "/chat",
-        json={
-            "documentId": uploaded_document,
-            "question": "What are the working hours?"
-        }
-    )
-
-    assert response.status_code == 200
-    data = response.json()
-
-    # Sources should include page numbers and text excerpts
-    for source in data["sources"]:
-        assert "page" in source
-        assert "text" in source
-        assert isinstance(source["page"], int)
-        assert len(source["text"]) > 0
-
-
-@pytest.mark.asyncio
-async def test_chat_invalid_document(client):
-    response = await client.post(
-        "/chat",
-        json={
-            "documentId": "nonexistent-id",
-            "question": "What is the policy?"
-        }
-    )
-
-    assert response.status_code == 404
-
-
-@pytest.mark.asyncio
-async def test_chat_empty_question(client, uploaded_document):
-    response = await client.post(
-        "/chat",
-        json={
-            "documentId": uploaded_document,
-            "question": ""
-        }
-    )
-
-    assert response.status_code == 400

package/refs/vbenchmark/tasks/ai-integration/rag-chatbot/support-bot/PROMPT.md

@@ -1,51 +0,0 @@
-# Task: Customer Support RAG Bot
-
-## Objective
-Build a customer support chatbot that answers questions using a knowledge base with RAG.
-
-## Requirements
-
-1. **Knowledge Base**
-   - Ingest FAQ documents
-   - Ingest help articles
-   - Ingest product documentation
-   - Support Markdown, PDF, HTML
-
-2. **Conversation**
-   - Multi-turn conversation
-   - Remember conversation context
-   - Clarifying questions
-   - Handoff to human when needed
-
-3. **Response Quality**
-   - Cite sources in responses
-   - Confidence threshold for answers
-   - "I don't know" for out-of-scope
-   - Suggested related articles
-
-4. **API**
-   - `POST /kb/ingest` - Add documents
-   - `POST /chat` - Send message
-   - `GET /chat/{session}/history` - Get history
-   - `POST /chat/{session}/feedback` - Rate response
-
-## Technical Stack
-- Python 3.11+
-- FastAPI
-- LangChain
-- ChromaDB or Pinecone
-- OpenAI/Anthropic
-
-## Files to Create
-- `app/kb/ingester.py` - Document ingestion
-- `app/kb/retriever.py` - RAG retrieval
-- `app/chat/bot.py` - Chatbot logic
-- `app/chat/memory.py` - Conversation memory
-- `app/routers/chat.py` - API endpoints
-
-## Success Criteria
-- [ ] Documents ingested and indexed
-- [ ] Relevant answers from knowledge base
-- [ ] Sources cited in responses
-- [ ] Conversation context maintained
-- [ ] Graceful handling of unknown queries

package/refs/vbenchmark/tasks/ai-integration/rag-chatbot/support-bot/task.yaml

@@ -1,24 +0,0 @@
-name: Customer Support RAG Bot
-category: ai-integration
-subcategory: rag-chatbot
-description: Build customer support chatbot with knowledge base RAG
-difficulty: medium
-stack: python
-timeout: 600
-tokenLimit: 150000
-
-tests:
-  functional: tests/test_support_bot.py
-
-tags:
-  - python
-  - rag
-  - chatbot
-  - support
-  - langchain
-
-evaluation:
-  functional: 0.5
-  quality: 0.2
-  cost: 0.15
-  speed: 0.15

package/refs/vbenchmark/tasks/ai-integration/structured-output/contract-analyzer/PROMPT.md

@@ -1,67 +0,0 @@
-# Task: Legal Contract Analyzer
-
-## Objective
-Build a contract analysis tool that extracts key terms, identifies risks, and summarizes legal documents.
-
-## Requirements
-
-1. **Document Processing**
-   - Accept PDF contracts
-   - Handle multi-page documents
-   - Extract text with section structure
-   - Identify document type (NDA, SLA, Employment)
-
-2. **Key Terms Extraction**
-   ```json
-   {
-     "parties": [{"name": "string", "role": "string"}],
-     "effectiveDate": "date",
-     "terminationDate": "date",
-     "renewalTerms": "string",
-     "paymentTerms": {
-       "amount": "number",
-       "currency": "string",
-       "schedule": "string"
-     },
-     "obligations": [{
-       "party": "string",
-       "description": "string",
-       "deadline": "string"
-     }],
-     "termination": {
-       "noticePeriod": "string",
-       "causes": ["string"]
-     }
-   }
-   ```
-
-3. **Risk Analysis**
-   - Identify unusual clauses
-   - Flag liability concerns
-   - Highlight missing standard clauses
-   - Compare against templates
-
-4. **API**
-   - `POST /analyze` - Upload contract
-   - Return extracted terms + risks
-   - `POST /compare` - Compare two contracts
-
-## Technical Stack
-- Python 3.11+
-- FastAPI
-- LangChain for chunking
-- OpenAI/Anthropic API
-
-## Files to Create
-- `app/services/extractor.py` - Document processing
-- `app/services/analyzer.py` - Term extraction
-- `app/services/risk_detector.py` - Risk analysis
-- `app/schemas/contract.py` - Data models
-- `app/routers/contracts.py` - API endpoints
-
-## Success Criteria
-- [ ] Parties identified correctly
-- [ ] Dates extracted accurately
-- [ ] Payment terms parsed
-- [ ] Risks flagged appropriately
-- [ ] Handles various contract types

package/refs/vbenchmark/tasks/ai-integration/structured-output/contract-analyzer/task.yaml

@@ -1,24 +0,0 @@
-name: Legal Contract Analyzer
-category: ai-integration
-subcategory: structured-output
-description: Analyze legal documents and extract key terms with Python
-difficulty: hard
-stack: python
-timeout: 900
-tokenLimit: 200000
-
-tests:
-  functional: tests/test_contract.py
-
-tags:
-  - python
-  - llm
-  - legal
-  - contracts
-  - analysis
-
-evaluation:
-  functional: 0.5
-  quality: 0.25
-  cost: 0.1
-  speed: 0.15

package/refs/vbenchmark/tasks/ai-integration/structured-output/invoice-parser/PROMPT.md

@@ -1,61 +0,0 @@
-# Invoice PDF Parser
-
-Build an invoice parser that extracts structured data from PDF invoices using an LLM.
-
-## Output Schema
-
-```python
-class LineItem(BaseModel):
-    description: str
-    quantity: float
-    unit_price: float
-    total: float
-
-class Invoice(BaseModel):
-    invoice_number: str
-    invoice_date: date
-    due_date: date | None
-    vendor_name: str
-    vendor_address: str | None
-    customer_name: str
-    customer_address: str | None
-    line_items: list[LineItem]
-    subtotal: float
-    tax_amount: float | None
-    tax_rate: float | None
-    total_amount: float
-    currency: str
-    payment_terms: str | None
-```
-
-## Requirements
-
-1. Use PyMuPDF or pdfplumber for PDF text extraction
-2. Use OpenAI or Anthropic with function calling for structured extraction
-3. Handle multi-page invoices
-4. Validate extracted data (totals should sum correctly)
-5. Return confidence scores for each field
-
-## API Endpoint
-
-```python
-@app.post("/parse")
-async def parse_invoice(file: UploadFile) -> Invoice:
-    ...
-```
-
-## Test Cases
-
-1. Simple single-page invoice
-2. Multi-page invoice with many line items
-3. Invoice with taxes and discounts
-4. Invoice in different currency (EUR, GBP)
-5. Poorly formatted/scanned invoice
-
-## Acceptance Criteria
-
-- [ ] Extracts all required fields
-- [ ] Line item totals sum to subtotal
-- [ ] Handles different date formats
-- [ ] Works with various invoice layouts
-- [ ] Returns validation errors for malformed invoices

package/refs/vbenchmark/tasks/ai-integration/structured-output/invoice-parser/task.yaml

@@ -1,27 +0,0 @@
-name: Invoice PDF Parser
-category: ai-integration
-subcategory: structured-output
-description: Extract structured data from invoice PDFs using LLM with function calling
-difficulty: hard
-stack: python
-timeout: 900
-tokenLimit: 200000
-
-tests:
-  functional: tests/parser.test.py
-
-docker:
-  compose: docker-compose.yaml
-
-tags:
-  - llm
-  - structured-output
-  - pdf
-  - extraction
-  - function-calling
-
-evaluation:
-  functional: 0.5
-  quality: 0.2
-  cost: 0.15
-  speed: 0.15

package/refs/vbenchmark/tasks/ai-integration/structured-output/receipt-scanner/PROMPT.md

@@ -1,65 +0,0 @@
-# Task: Receipt Scanner with OCR + LLM
-
-## Objective
-Build a receipt scanner that extracts structured data from receipt images using OCR and LLM.
-
-## Requirements
-
-1. **Image Processing**
-   - Accept JPEG, PNG images
-   - Handle rotated/skewed images
-   - Support various receipt formats
-   - Multi-receipt in one image
-
-2. **Data Extraction**
-   ```typescript
-   interface Receipt {
-     merchant: {
-       name: string;
-       address?: string;
-       phone?: string;
-     };
-     date: string; // ISO 8601
-     items: Array<{
-       name: string;
-       quantity: number;
-       unitPrice: number;
-       totalPrice: number;
-     }>;
-     subtotal: number;
-     tax: number;
-     total: number;
-     paymentMethod?: string;
-   }
-   ```
-
-3. **LLM Integration**
-   - Use vision model (GPT-4V, Claude Vision)
-   - Or OCR + text LLM pipeline
-   - Validate extracted amounts (sum check)
-   - Handle unclear/damaged receipts
-
-4. **API**
-   - `POST /scan` - Upload receipt image
-   - Return structured receipt data
-   - Include confidence score
-
-## Technical Stack
-- TypeScript/Node.js
-- Express or Hono
-- OpenAI Vision API or Tesseract + LLM
-- Sharp for image preprocessing
-
-## Files to Create
-- `src/routes/scan.ts` - API endpoint
-- `src/services/ocr.ts` - OCR processing
-- `src/services/parser.ts` - LLM parsing
-- `src/services/validator.ts` - Data validation
-- `src/types/receipt.ts` - Type definitions
-
-## Success Criteria
-- [ ] Receipt images processed correctly
-- [ ] All line items extracted
-- [ ] Totals match sum of items
-- [ ] Date parsed correctly
-- [ ] Handles various receipt formats

package/refs/vbenchmark/tasks/ai-integration/structured-output/receipt-scanner/task.yaml

@@ -1,24 +0,0 @@
-name: Receipt Scanner with OCR + LLM
-category: ai-integration
-subcategory: structured-output
-description: Extract receipt data using OCR and LLM in TypeScript
-difficulty: medium
-stack: nodejs
-timeout: 600
-tokenLimit: 150000
-
-tests:
-  functional: tests/receipt.test.ts
-
-tags:
-  - typescript
-  - ocr
-  - llm
-  - receipt
-  - vision
-
-evaluation:
-  functional: 0.5
-  quality: 0.2
-  cost: 0.15
-  speed: 0.15

package/refs/vbenchmark/tasks/ai-integration/structured-output/resume-parser/PROMPT.md

@@ -1,70 +0,0 @@
-# Task: Resume Parser with LLM
-
-## Objective
-Build a resume parser that extracts structured data from PDF/DOCX resumes using LLM with function calling.
-
-## Requirements
-
-1. **File Processing**
-   - Accept PDF and DOCX files
-   - Extract text with formatting hints
-   - Handle multi-page resumes
-   - Support multiple languages
-
-2. **Data Extraction**
-   ```json
-   {
-     "name": "string",
-     "email": "string",
-     "phone": "string",
-     "location": "string",
-     "summary": "string",
-     "experience": [{
-       "company": "string",
-       "title": "string",
-       "startDate": "YYYY-MM",
-       "endDate": "YYYY-MM | Present",
-       "description": "string",
-       "highlights": ["string"]
-     }],
-     "education": [{
-       "institution": "string",
-       "degree": "string",
-       "field": "string",
-       "graduationDate": "YYYY-MM"
-     }],
-     "skills": ["string"],
-     "certifications": ["string"]
-   }
-   ```
-
-3. **LLM Integration**
-   - Use function calling for structured output
-   - Validate output against schema
-   - Handle ambiguous data gracefully
-   - Confidence scores for uncertain fields
-
-4. **API**
-   - `POST /parse` - Upload and parse resume
-   - Return structured JSON
-
-## Technical Stack
-- Python 3.11+
-- FastAPI
-- OpenAI/Anthropic API
-- PyMuPDF for PDF
-- python-docx for DOCX
-
-## Files to Create
-- `app/main.py` - FastAPI app
-- `app/services/extractor.py` - Text extraction
-- `app/services/parser.py` - LLM parsing
-- `app/schemas/resume.py` - Pydantic models
-- `app/routers/parse.py` - API endpoint
-
-## Success Criteria
-- [ ] PDF and DOCX files parsed
-- [ ] All fields extracted correctly
-- [ ] Dates normalized to YYYY-MM
-- [ ] Skills list is comprehensive
-- [ ] Output validates against schema

package/refs/vbenchmark/tasks/ai-integration/structured-output/resume-parser/task.yaml

@@ -1,24 +0,0 @@
-name: Resume Parser with LLM
-category: ai-integration
-subcategory: structured-output
-description: Extract structured data from resumes using LLM
-difficulty: medium
-stack: python
-timeout: 600
-tokenLimit: 150000
-
-tests:
-  functional: tests/test_resume_parser.py
-
-tags:
-  - python
-  - llm
-  - parsing
-  - resume
-  - structured-output
-
-evaluation:
-  functional: 0.5
-  quality: 0.2
-  cost: 0.15
-  speed: 0.15

package/refs/vbenchmark/tasks/api-integrations/advanced/api-analytics/PROMPT.md

@@ -1,15 +0,0 @@
-# API Analytics Platform
-
-Build comprehensive API analytics system.
-
-## Requirements
-1. Real-time request metrics
-2. Usage pattern analysis
-3. Anomaly detection (ML-based)
-4. Customer usage dashboards
-5. Endpoint performance tracking
-6. Error rate monitoring
-7. Geographic distribution
-8. API versioning analytics
-9. Custom event tracking
-10. Export and reporting

package/refs/vbenchmark/tasks/api-integrations/advanced/api-analytics/task.yaml

@@ -1,16 +0,0 @@
-name: API Analytics Platform
-category: api-integrations
-subcategory: advanced
-description: Build API analytics with real-time metrics, usage patterns, and anomaly detection
-difficulty: hard
-stack: go
-timeout: 1200
-tokenLimit: 250000
-tests:
-  functional: tests/analytics_test.go
-tags: [go, analytics, clickhouse, metrics, anomaly]
-evaluation:
-  functional: 0.5
-  quality: 0.25
-  cost: 0.1
-  speed: 0.15

package/refs/vbenchmark/tasks/api-integrations/advanced/api-gateway/PROMPT.md

@@ -1,15 +0,0 @@
-# API Gateway with Plugins
-
-Build Kong/Envoy-style API gateway.
-
-## Requirements
-1. Plugin architecture (auth, transform, logging)
-2. Request/response transformation
-3. Traffic splitting and mirroring
-4. Service discovery integration
-5. Dynamic configuration
-6. mTLS support
-7. Request buffering/streaming
-8. Admin API
-9. Plugin SDK
-10. Hot reload without downtime

package/refs/vbenchmark/tasks/api-integrations/advanced/api-gateway/task.yaml

@@ -1,16 +0,0 @@
-name: API Gateway with Plugins
-category: api-integrations
-subcategory: advanced
-description: Build extensible API gateway with plugin system, request transformation, and traffic management
-difficulty: hard
-stack: go
-timeout: 1200
-tokenLimit: 250000
-tests:
-  functional: tests/gateway_test.go
-tags: [go, api-gateway, plugins, proxy, traffic]
-evaluation:
-  functional: 0.5
-  quality: 0.25
-  cost: 0.1
-  speed: 0.15

package/refs/vbenchmark/tasks/api-integrations/advanced/api-mocking/PROMPT.md

@@ -1,15 +0,0 @@
-# Dynamic API Mocking Service
-
-Build WireMock-style API mocking service.
-
-## Requirements
-1. Dynamic response generation
-2. Stateful scenarios (sequences)
-3. Request matching (body, headers, query)
-4. Fault injection (delays, errors)
-5. Record and playback
-6. Contract verification
-7. Template responses (Handlebars)
-8. Webhook simulation
-9. Admin API for configuration
-10. Proxy mode with recording

package/refs/vbenchmark/tasks/api-integrations/advanced/api-mocking/task.yaml

@@ -1,16 +0,0 @@
-name: Dynamic API Mocking Service
-category: api-integrations
-subcategory: advanced
-description: Build API mocking with dynamic responses, stateful scenarios, and contract verification
-difficulty: hard
-stack: typescript
-timeout: 1200
-tokenLimit: 250000
-tests:
-  functional: tests/mock.test.ts
-tags: [typescript, mocking, testing, openapi, wiremock]
-evaluation:
-  functional: 0.5
-  quality: 0.25
-  cost: 0.1
-  speed: 0.15