Skip to content

docent_chrome_plugin

Chrome extension page guide whose small typed decisions (request type, element to click, match or not) use Jev or local Laya.

Projectbeta

Overview

Docent is a Chrome extension that acts as a guide for the current page. You can ask questions about the page, have it summarised or read aloud, turn it into a two-voice podcast, highlight or hide things by meaning, restyle the page, or let it click and type for you, by text, microphone or hands-free conversation.

A decision model makes Docent's small typed decisions: what kind of request you made, which element to click, whether a post matches, yes or no. You choose Jev, which runs on TypeSafe's API with page text sent there and is described as about a second per step and most accurate, or Laya, which runs in the browser after a one-time 290 or 440 MB download, sends nothing out and is better at yes/no and find/hide than multi-step tasks. Summaries and voice features use LiveKit separately.

What the decision model decides

TypeQuestionAnswers
ChoiceWhat kind of request did the user make?—
ChoiceWhich page element should be clicked?—
Yes / noDoes this post match the user's description?yes · no

How it works

  1. User asks Docent something about the page
  2. Decision model classifies the request type
  3. For actions, it picks the element to act on; for filters, it judges each item
  4. Extension performs the click, highlight or hide

Strengths

  • Local Laya option keeps page text in the browser
  • Jev option for higher accuracy
  • Many page tools in one extension

Caveats

  • Installed unpacked in developer mode
  • Jev sends page text to TypeSafe
  • Laya is weaker on multi-step tasks

Getting started

Load the folder unpacked at chrome://extensions, open Docent with Alt+J, and choose Jev or Laya in settings.

At a glance

Type
Project
Status
beta
Language
JavaScript
Runtime
Chrome extension
Has code
yes

Use cases and integrations

How curated lists describe it

Frequently asked questions

Do I need a TypeSafe key?

Only for Jev; Laya runs locally in the browser without a key.

What does the decision model do?

It classifies requests, picks elements to click, and judges whether items match.

Is page text sent anywhere?

With Jev it goes to TypeSafe; with Laya it stays in the browser.

Related projects

  • Chrome extension adding a second Twitch chat column that shows only messages Jev classifies as helpful, questions, funny or feedback.

    Browser extensionbetaJavaScript
  • Chrome extension that uses Jev to spot native ads and sponsored feed cards, alongside network and media ad blocking.

    Browser extensionbetaJavaScript
  • Chrome extension tagging each X post's intent (inform, persuade, provoke, sell) plus rage-bait and hidden-ad flags via Jev.

    ProjectbetaJavaScript
  • Node app that controls Chromium by voice, asking Jev a dozen typed questions per partial transcript to act before you finish.

    ProjectprototypeJavaScript
  • Zero-build local playground for Jev with browser-agent, intent routing and context-pruning presets, including a Playwright agent.

    AppbetaJavaScript
  • Privacy-first X filtering extension with local keyword and account rules, plus an optional Jev second pass on replies no rule matched.

    Pluginproduction