@openruntime/extension-imitate
v0.1.7
Published
This OpenRuntime Extension records a real browser walkthrough and turns the captured interactions, page context, Runtime state, and optional spoken intent into a reusable JavaScript script draft.
Downloads
650
Readme
@openruntime/extension-imitate
This OpenRuntime Extension records a real browser walkthrough and turns the captured interactions, page context, Runtime state, and optional spoken intent into a reusable JavaScript script draft.
Install
openruntime extensions add @openruntime/extension-imitateRecord a manual walkthrough
Prepare the recording first so browser interaction and optional microphone capture are active from page startup. The output defaults to ./recordings:
openruntime record start --micThen open the page through OpenRuntime. The CLI starts and injects the Bridge while the recording Extension injects its page capture script into the same browser launch:
openruntime open https://example.com/ --uiThe guided workflow can use about:blank when no URL is provided, and the output path is optional. When the walkthrough is finished, stop the recording with the output path returned by start:
openruntime record stop --out ./recordings/openruntime-<timestamp>.orrecStopping captures final state and writes generated-script.mjs by default. It leaves the current page open; close it through the normal page lifecycle when the workflow is complete:
openruntime stopRecording preparation refuses to replace an already open page. Close the current page first, prepare the recording, and then open the page to record. Stopping refuses to mix evidence if another openruntime open replaced the recorded page.
Regenerate or transcribe
Regenerate a script from an existing recording:
openruntime record generate-script \
--input ./recordings/example.orrec \
--out ./scripts/example.mjsWhen microphone audio exists, create timestamped text with:
OPENAI_API_KEY=... openruntime record transcribe \
--input ./recordings/example.orrecThe default transcription model is whisper-1; use --model to select another compatible model. Do not treat a script that only opens the last URL as complete when recorded audio has not been transcribed.
Fixed-duration capture
For an unattended time-bounded recording:
openruntime open https://example.com/ --ui
openruntime record \
--out ./recordings/example.orrec \
--duration 30000
openruntime stopFor an Agent-guided installation and workflow, see the English guide or 中文指南.
