EchoFill is a dependency-free Chrome Manifest V3 extension that turns a spoken conversation into values for the form on the active page.
- Click the EchoFill toolbar icon. EchoFill explicitly opens its side panel for the clicked tab; this click also activates the scoped
activeTabaccess used to read that tab. - Click Start recording and speak naturally. EchoFill uses Chrome
SpeechRecognition, preferring on-device conversation-quality recognition when the local language pack is available. - Click Finish recording. EchoFill reads the visible form fields on the current
http(s)page and asks Chrome's built-inLanguageModelPrompt API for structured field matches. - Review the matches, uncheck any field you do not want to fill, and click Apply selected. Existing values are preserved unless Overwrite existing values is enabled.
If you navigate to another page while the panel remains open, click the EchoFill toolbar icon again on the new page before recording so the new tab receives the temporary activeTab grant.
On the first recording, Chrome may open a short EchoFill permission page because Chrome can suppress microphone prompts inside a side panel. Click Grant microphone access there; the page closes after permission is granted.
- Open
chrome://extensionsin Chrome. - Enable Developer mode.
- Click Load unpacked and select this directory.
- Pin EchoFill, open a normal
http(s)page with a form, and click the toolbar icon.
The first use of Chrome built-in AI may need to download its on-device model. Chrome's built-in AI availability depends on the user's Chrome channel, device, language, and administrator settings. EchoFill does not include a cloud AI fallback or an API key.
npm run checkThe extension intentionally has no runtime dependencies or build step. It uses activeTab and scripting so page access is scoped to the tab where the user invokes the extension. Restricted pages such as chrome:// and Chrome Web Store pages cannot be inspected.