Input guard

Evaluates inbound text before you call your LLM. The handler runs the full input pipeline (normalization, heuristics, session layer, classifier, and configured provider/modules).

For external providers (claude / openai), decisions are module-driven: flag/block require module-attributed modules_triggered entries; otherwise the decision resolves to pass.

Request

Response

Examples

curl -s -X POST http://localhost:8000/v1/guard/input \
  -H "Content-Type: application/json" \
  -d '{"text":"Hello"}'

import httpx

r = httpx.post(
    "http://localhost:8000/v1/guard/input",
    json={"text": "Hello"},
)
print(r.json())

See also

• Output guard
• Sessions