Skip to main content

verdictan init

Create a Verdictan policy project with a starter policy-config.yaml and test files.

Usage​

verdictan init [OPTIONS]

Options​

FlagDescription
--listList available template IDs and exit
--template <id>Use an industry template as starting point
--dir <path>Directory to initialize (default: working directory)
--forceOverwrite generated files in the target directory
--config <path>Optional CLI config file to use when fetching templates from the API
--api-url <url>Override the Verdictan API URL for template fetches
--profile <name>CLI profile to use when fetching templates (default: default)

Template fetches authenticate through VERDICTAN_API_TOKEN or a stored profile from verdictan auth login. The CLI no longer accepts API tokens as command-line flags.

Examples​

List available templates​

verdictan init --list

With a token and an available templates endpoint, output contains each template ID, name, type, and category. Without a token, the command lists template IDs from a nearby repository demos/ directory. It also uses this local list when it cannot use the API response. A standalone installation can show no rows when local demos and the API catalog are unavailable.

verdictan init --list excludes these removed solution identifiers:

  • payment-processing
  • credit-scoring
  • tax-advisory
  • tax-authority

verdictan init --template fails for a removed identifier. The command does not select a retained template or write a default pack.

Use the default files​

verdictan init

The command creates these files:

./
├── policy-config.yaml
└── tests/
    └── blocks_obvious_injection.json

Generated policy-config.yaml:

pack:
  name: my-policy
  version: 0.1.0
  enabled: true
  description: "Verdictan policy configuration"

policies:
  chain:
    - prompt-injection

policy:
  prompt-injection: {}

Use a template​

Initialize with an industry template as the starting point:

verdictan init --template finance --dir ./finance-gateway
verdictan init --template healthcare-us-hipaa --dir ./hipaa-gateway
verdictan init --template colorado-ai-act --dir ./colorado-ai-gateway
verdictan init --template third-party-ai-risk --dir ./vendor-review-gateway

Continue the setup​

After verdictan init, follow this sequence:

cd ./finance-gateway

# 1. Review and customize the config
vim policy-config.yaml

# 2. Add a provider target and reference its credential from the environment
#    or secret store. Use a provider and model your account supports.

# 3. Validate
verdictan policy lint --file policy-config.yaml

# 4. Run tests
verdictan policy test --json

# 5. Start the gateway from the declarative config
verdictan gateway run \
  --listen 127.0.0.1:41002 \
  --agent docs-demo \
  --policy-config policy-config.yaml

The default scaffold contains a policy and one golden test, but no provider target. Before you route model traffic, add a provider. Validate the provider. A connected gateway also must have VERDICTAN_API_TOKEN. This runtime credential is different from the application token that calls the gateway.

The template loader verifies that returned starter content is YAML. verdictan policy lint validates the schema and runtime mode. After initialization, do not skip it.

For the broader operating model, read Config-First Workflow.

Next steps​