JEV × SOCAI

Social research
in your real browser.

Describe an Instagram, TikTok, or LinkedIn research goal. TypeSafe Jev chooses the next bounded operation; the local socai CLI executes it in Chrome; captured evidence streams into cards and a cited report.

Local UI · real CLI calls · no mock fallback · MIT licensed

One goal. Visible evidence.

The recording is at 1× speed. Current builds also show every operation Jev chooses.

Jev Social local demo routing a goal to socai and streaming captured post cards Open the local setup guide ↗

Jev decides. socai acts.

The model never writes shell commands or arbitrary selectors.

  1. 01

    Route the goal

    Jev selects Instagram, TikTok, or LinkedIn from a typed choice.

  2. 02

    Choose one operation

    Search, open an observed post or profile, read comments, download selected video, or finish.

  3. 03

    Run the real CLI

    socai drives your existing Chrome session and returns structured evidence.

  4. 04

    Stream the result

    Post cards stay visible while Jev evaluates the new state and builds the final report.

A changing decision space, not a blank check.

Every step offers only operations supported by the installed socai CLI and only targets observed in captured results. Attempted actions disappear from the next choice set.

  • Low-confidence and malformed decisions do not execute.
  • Login gates and step limits preserve honest partial results.
  • Sources, comments, media, and timing remain inspectable.
jev-social
goal  find emerging design creators

jev   instagram.search
socai captured 4 post cards

jev   instagram.read_post [2]
socai captured caption + comments

jev   finish
done  source-linked report

Your browser. Your session. Your evidence.

Requires Node 20+ and an OpenRouter key with Jev access. Onboarding offers to install the current socai CLI when it is missing.

TERMINAL
npx --yes github:socai-io/jev-social onboard
npx --yes github:socai-io/jev-social

Social research should show its work.