Command line reference

Usage:
  dummyrun run --cast DIR --persona ID --scenario ID --target CONFIG --out DIR \
      --browser chrome|firefox --provider PROVIDER --model MODEL [--region REGION] [--endpoint URL] \
      [--api-key-env NAME] [--project ID] [--reasoning-effort LEVEL] [--openai-api chat|responses]
  dummyrun a11y [--sitemap URL] [URL...] [--browser chrome|firefox] [--allow FILE] [--json FILE]
  dummyrun version

Run "dummyrun run -h" or "dummyrun a11y -h" for every flag. "dummyrun audit a11y" is the same as "dummyrun a11y".

Providers:
  azure              Azure OpenAI. --endpoint or DUMMYRUN_AZURE_ENDPOINT; sign in with "az login" or another
                     DefaultAzureCredential source. --region is required.
  openai-compatible  Any OpenAI-style service or local server. --endpoint is the base URL, such as
                     https://api.openai.com/v1 or http://localhost:11434/v1. --api-key-env names the variable
                     that holds the key (leave it out for a local server). --region is required (use "local").
  anthropic          Anthropic Messages API. The key is read from ANTHROPIC_API_KEY or --api-key-env.
  gemini             Gemini Developer API. The key is read from GEMINI_API_KEY or --api-key-env.
  vertex             Gemini on Vertex AI with Application Default Credentials. --project (or
                     GOOGLE_CLOUD_PROJECT) and --region are required.
  bedrock            Amazon Bedrock Converse with the AWS default credential chain. --region defaults to eu-west-2.

An API key is read from the environment when the run starts. It is never printed or written to a run record.

Usage of run:
  -api-key-env string
    	name of the environment variable that holds the API key
  -api-version string
    	Azure OpenAI API version (default 2025-04-01-preview)
  -batch string
    	batch ID grouping runs started together
  -browser string
    	chrome or firefox (default "chrome")
  -cast string
    	cast directory
  -endpoint string
    	base URL of the provider (Azure endpoint, or the OpenAI-compatible base URL)
  -headful
    	show the browser window
  -model string
    	deployment or model name
  -openai-api string
    	OpenAI-style API: chat (default) or responses, for models that need it to call tools
  -out string
    	directory for run records
  -persona string
    	persona ID
  -project string
    	Google Cloud project for the vertex provider
  -provider string
    	model provider: azure, openai-compatible, anthropic, gemini, vertex, bedrock (default "azure")
  -reasoning-effort string
    	reasoning effort for OpenAI-style models: none, minimal, low, medium, high or xhigh
  -region string
    	region the model runs in, such as uksouth or global (optional where the provider has a default)
  -scenario string
    	scenario ID
  -steps int
    	step budget override (up to 40)
  -target string
    	target configuration (JSON)
  -tokens int
    	token budget (input plus output)

Usage:
  dummyrun a11y [--sitemap URL] [--rebase URL] [URL...] [flags]

Runs the axe-core WCAG 2.2 A and AA rules on each page in a real browser. No model is called.

Flags:
  --sitemap URL     a sitemap (or sitemap index) listing the pages to test
  --rebase URL      move every sitemap URL onto this origin, to test a local copy of a site
  --browser NAME    chrome or firefox (default chrome)
  --viewport NAME   desktop (1280 wide), phone (375 wide) or both (default both)
  --scheme NAME     light, dark or both (default both)
  --allow FILE      accepted violations, one per line: rule | page | selector | reason
  --json FILE       write the full report as JSON; "-" writes it to standard output
  --headful         show the browser window

Exit status: 0 when nothing fails, 1 when a violation is not in the allow-list or a page cannot be
tested, 2 for a usage error.

Usage:
  dummyrun-otel export [flags] FILE|DIR...   send run records to an OTLP endpoint
  dummyrun-otel watch  [flags] DIR           send new run records as they appear
  dummyrun-otel version

The OTLP endpoint, headers and TLS settings come from the standard variables:
OTEL_EXPORTER_OTLP_ENDPOINT, OTEL_EXPORTER_OTLP_HEADERS, OTEL_EXPORTER_OTLP_PROTOCOL,
OTEL_RESOURCE_ATTRIBUTES. Grafana annotations need --grafana-url and GRAFANA_TOKEN.
Run "dummyrun-otel export -h" for the flags.

Usage of dummyrun-otel export:
  -dry-run
    	print what would be sent as JSON and send nothing
  -grafana-url string
    	Grafana base URL for annotations; the token comes from GRAFANA_TOKEN
  -logs-max-age duration
    	skip log records older than this, because Loki rejects them by default; 0 sends every record (default 168h0m0s)
  -metrics-temporality string
    	metric temporality: cumulative or delta (default "cumulative")
  -protocol string
    	OTLP transport: http/protobuf (default) or grpc
  -timeout duration
    	how long to wait for the final flush (default 30s)
  -verify-loki string
    	Loki base URL; after the export, count the log lines back per run and fail when any are missing. Basic auth comes from LOKI_BASIC_AUTH as user:password

Usage of dummyrun-otel watch:
  -dry-run
    	print what would be sent as JSON and send nothing
  -existing
    	also send the records already in the directory at start
  -grafana-url string
    	Grafana base URL for annotations; the token comes from GRAFANA_TOKEN
  -interval duration
    	how often to look for new files (default 5s)
  -logs-max-age duration
    	skip log records older than this, because Loki rejects them by default; 0 sends every record (default 168h0m0s)
  -metrics-temporality string
    	metric temporality: cumulative or delta (default "cumulative")
  -protocol string
    	OTLP transport: http/protobuf (default) or grpc
  -timeout duration
    	how long to wait for the final flush (default 30s)
  -verify-loki string
    	Loki base URL; after the export, count the log lines back per run and fail when any are missing. Basic auth comes from LOKI_BASIC_AUTH as user:password