Bridge Application

Command Line

Started with arguments, the bridge executes one run and exits — no web server, no MongoDB, no Elasticsearch. Nothing is persisted, so the CLI is for interactive work: reproducing a failing conversation, watching a device while it is driven, or trying a configuration before saving it.

./gradlew :promethist-bridge:bootJar

java -jar bridge/build/libs/promethist-bridge-*.jar \
  --key=agent:tester \
  --device-uri="https://eu.promethist.ai?key=agent:acme-support" \
  --timeout=300 \
  --playback

Options

Option Default Meaning

Option	Default	Meaning
`--key`	required	Pipeline key of the tester agent.
`--device-uri`	required	What the tested application runs on — see Devices.
`--engine-url`	`https://eu.promethist.ai`	Engine the tester connects to.
`--client-url`	—	URL the device should open, typically an agent link.
`--relay-url`	—	Relay service, for device types that need it.
`--device-query`	—	Extra query string appended to the device URI. Repeat the option to produce several runs.
`--device-ref`	`bridge:{host MAC}`	Device ref reported to the engine.
`--locale`	`en-US`	Conversation locale.
`--sample-rate`	platform default	Audio sample rate.
`--audio-speed`	`1`	Pacing multiplier for audio sent to the device.
`--text-only`	`false`	Run the conversation as text only.
`--no-system-prompt`	`false`	Suppress the engine’s system prompt for the tester.
`--speech-processing`	`false`	Enable server-side speech processing.
`--timeout`	`600`	Overall run timeout in seconds.
`--max-turns-per-session`	—	Turn cap per session.
`--session-count`	`1`	Consecutive conversations per run.
`--device-count`	`1`	Devices to run in parallel.
`--device-user`, `--device-secret`	—	Device provider credentials, required for `bs://` runs.
`--device-delay`	`5000`	Milliseconds between starting devices.
`--parallelism`	`1`	Concurrent runs.
`--playback`	`false`	Show the device screen in a window and play the audio locally. CLI mode only — the web service runs headless.
`--video-file`	—	Write the recording to this local path; `%id` is replaced by the run ID.

--key

required

Pipeline key of the tester agent.

--device-uri

required

What the tested application runs on — see Devices.

--engine-url

https://eu.promethist.ai

Engine the tester connects to.

--client-url

—

URL the device should open, typically an agent link.

--relay-url

—

Relay service, for device types that need it.

--device-query

—

Extra query string appended to the device URI. Repeat the option to produce several runs.

--device-ref

bridge:{host MAC}

Device ref reported to the engine.

--locale

en-US

Conversation locale.

--sample-rate

platform default

Audio sample rate.

--audio-speed

1

Pacing multiplier for audio sent to the device.

--text-only

false

Run the conversation as text only.

--no-system-prompt

false

Suppress the engine’s system prompt for the tester.

--speech-processing

false

Enable server-side speech processing.

--timeout

600

Overall run timeout in seconds.

--max-turns-per-session

—

Turn cap per session.

--session-count

1

Consecutive conversations per run.

--device-count

1

Devices to run in parallel.

--device-user, --device-secret

—

Device provider credentials, required for bs:// runs.

--device-delay

5000

Milliseconds between starting devices.

--parallelism

1

Concurrent runs.

--playback

false

Show the device screen in a window and play the audio locally. CLI mode only — the web service runs headless.

--video-file

—

Write the recording to this local path; %id is replaced by the run ID.

Variable substitution is not applied in CLI mode — pass literal values.

Reading the Outcome

Run IDs and progress go to standard output, and each run logs its final status and reason on completion. Since nothing is stored, note the tester’s session ID from the log if the conversation needs to be inspected in the engine afterwards.

For repeatable, recorded and evaluated runs use the REST API instead — that is the mode which stores results and fetches evaluations.