Guardrails

How guardrails work.

Four pre-call checks and two post-call checks run on every request. Blocked requests return 403 — never enqueued, never billed.

Pipeline order

Guardrails always run before routing

→ Auth + rate limit check

→ [PRE-CALL GUARDRAILS]

1. Prompt injection detection (ML classifier, ~30ms)

2. PII detection (~20ms)

3. Content moderation (keyword + ML, ~5ms)

4. Tool-call policy check (allowlist/denylist, ~10ms)

↳ blocked? Return 403. Request never reaches LLM. Never billed.

→ Semantic cache check → Routing → LLM call

→ [POST-CALL GUARDRAILS]

5. Response content filter

6. Response PII scan

→ Token metering → Response to client

Guardrail types

Four pre-call layers

Prompt Injection

ML-based prompt-injection classifier. Threshold: 0.85 — configurable per customer. ~30ms CPU latency. Cannot be disabled system-wide; threshold is adjustable.

PII Masking

PII detection engine. Entities: EMAIL, PHONE, CREDIT_CARD, SG NRIC, ID NIK, IP_ADDRESS, URL. Four modes: redact, tokenize, synthetic, block.

Content Moderation

Keyword + ML classifier. English + Bahasa Indonesia. ~5ms latency. Configurable per policy group.

Conversational Guardrails

YAML + flow-rule config per policy group. Conversational flow enforcement — blocks topic steering, jailbreaks, system-prompt disclosure.

PII modes

Four ways to handle detected PII

// Input: "My email is [email protected]"
// Output: "My email is [REDACTED]"

// API key setting:
{ "pii_mode": "redact" }

Per-key overrides

Override guardrail defaults per API key

// PATCH /anoman/v1/keys/{id}
{
  "guardrail_overrides": {
    "injectionEnabled": false,
    "piiEnabled": true,
    "contentModerationEnabled": false
  }
}

// Priority stack (highest → lowest):
// 1. System forced (ops emergency)
// 2. Per-key guardrail_overrides
// 3. Policy group defaults
// 4. System defaults

Conversational Guardrails

Conversational flow enforcement

Configure via the dashboard Guardrails tab. YAML config + flow rules are stored per policy group and cached by config hash for performance.

# Policy group conversational-guardrail config
models:
  - type: main
    engine: openai
    model: gpt-4o

rails:
  input:
    flows:
      - self check input
  output:
    flows:
      - self check output

Full guardrails overview: Platform / Guardrails →

Secure every AI call with guardrails.

Get API Key API Reference