Full Request Lifecycle

This walkthrough shows a complete request from user input to validated response, hitting every module in the pipeline.

Scenario

A customer asks: "What is the refund policy?" using the RAG_QA prompt package.

Step 1: Evaluate Input

curl -X POST http://localhost:8000/v1/guardrails/evaluate-input \
  -H "Content-Type: application/json" \
  -H "X-Tenant-ID: default" \
  -d '{"text": "What is the refund policy?"}'

Response: decision: "allow" -- all 7 checks pass.

Step 2: Search for Grounding Context

curl -X POST http://localhost:8000/v1/retrieval/search \
  -H "Content-Type: application/json" \
  -d '{"query": "refund policy", "require_grounding": true}'

Response: Citations found with confidence scores. grounded: true.

Step 3: Compile Prompt

curl -X POST http://localhost:8000/v1/prompts/compile \
  -H "Content-Type: application/json" \
  -d '{
    "package_name": "rag_qa",
    "user_message": "What is the refund policy?",
    "retrieved_context": "[1] Company Refund Policy (confidence: 0.667)\nRefunds are available within 30 days of purchase with a valid receipt."
  }'

Response: Compiled messages with system instructions, grounding, and refusal policy. Lint warnings (if any).

Step 4: Call LLM (via Gateway)

The compiled messages are sent to the LLM provider through the gateway.

Step 5: Validate Output

curl -X POST http://localhost:8000/v1/outputs/validate \
  -H "Content-Type: application/json" \
  -d '{
    "output_text": "Based on the company policy, refunds are available within 30 days of purchase with a valid receipt [1].",
    "context_text": "Refunds are available within 30 days of purchase with a valid receipt.",
    "require_citations": true
  }'

Response: decision: "pass" -- all checks pass, citations present, no hallucination.

Step 6: Evaluate Policy

curl -X POST http://localhost:8000/v1/policies/evaluate \
  -H "Content-Type: application/json" \
  -d '{
    "policy_name": "default",
    "context": {"requires_grounding": true, "grounded": true}
  }'

Response: decision: "allow" -- grounding requirement satisfied.

Result

The validated, grounded, policy-compliant response is returned to the user.