Configuration & Policy Overview

This section shows how declarative config controls gateway behavior. Use it to select policy types and set their sequence in policies.chain. It also shows runtime behavior.

Start here

Your task	Start with
Understand the supported top-level schema	Declarative Config Reference
Browse supported controls before selecting one	Policy Controls Catalog
Define provider targets, routing, or fallback	Providers Configuration and data-routing-policy
Add tests before rollout	Config Testing and verdictan policy test
Protect sensitive or controlled data	pii-detector and dlp-filter
Defend the input boundary	prompt-injection

Follow this sequence

1. Read the Declarative Config Reference to select the correct document shape.
2. Select policy types from the Policy Controls Catalog.
3. Add them to policies.chain and tune each block in policy.<kind>.
4. Validate the config with lint and tests before rollout.
5. Add the audit marker.
6. Verify request results with verdictan events or the Events API.
7. Add History only when capture is enabled and stored session context is necessary.

How policies work

Primary policy groups

Request and routing controls

Policy type	Purpose
prompt-injection	Block prompt-injection and jailbreak signals before upstream calls
pii-detector	Detect sensitive identifiers and redact or block request content
dlp-filter	Apply broader pattern-based filtering for sensitive or proprietary data
rbac	Apply role, identity, and sensitivity-based access rules
agent-firewall	Limit agent tool use and dangerous action classes
data-routing-policy	Filter provider targets by retention or training metadata

Output controls

Policy type	Purpose
quality-scorer	Score output quality and block or replace failed output
human-oversight	Return an escalated result and do not deliver output
citation-verifier	Verify groundedness against context

Audit and evidence

Policy type	Purpose
audit-logger	Emit an allow-only audit marker. Configure retention and storage through their owning workflows.

Example configuration

pack:
  name: "my-project"
  version: "0.1.0"
  enabled: true

policies:
  chain:
    - audit-logger
    - prompt-injection
    - pii-detector
    - quality-scorer

policy:
  prompt-injection:
    attack_patterns:
      - "ignore.*previous.*instructions"
      - "forget.*system.*prompt"
    encoding:
      decode_base64: true
      normalize_unicode: true
      detect_homoglyphs: true
    boundaries:
      enforce_delimiters: true
      reject_fake_boundaries: true
    response:
      action: block

  pii-detector:
    action: redact
    pci_mode: true

  quality-scorer:
    assertions:
      - type: word-count
        config:
          min: 20
    thresholds:
      min_aggregate: 0.7

  audit-logger: {}

Key rules

• Chain sequence: Policies run in the sequence in policies.chain.
• First block: After the gateway blocks a turn, subsequent policies do not run for that stage.
• Audit marker position: Put audit-logger before blocking controls when each evaluated request must contain the marker.
• Redaction and buffering: Some policies redact request text and also buffer and examine output.
• Phase limits: human-oversight runs only on output. language-validator enforces input checks only.

See the Declarative Config Reference for the complete schema and the Policy Controls Catalog for the inventory of supported controls.

Next steps

• Declarative Config Reference — Full config schema
• Policy Controls Catalog — Browse policy types and starter snippets
• Config-First Workflow — Operating model for policy deployment
• Config Testing — validate the specified chain before rollout