Creators, posts, Reels, comments
Search a topic, inspect an observed profile and its post cards, then open a relevant post or Reel and read its comments.
- Starts with
- Search
- Detail
- Post + comments
SOCIAL RESEARCH WITH JEV
Jev for social research.
Evidence attached.
Jev Social is an open-source, read-only social media research agent for Instagram, TikTok, and LinkedIn. Jev chooses what to inspect next; the local socai CLI performs that operation in Chrome; captured sources stay beside the report.
Instagram
TikTok
LinkedIn
Read-only operations · observed targets · partial results stay honest · MIT licensed
WHAT DOES IT DO, AND WHERE DOES JEV FIT?
A decision loop around real evidence.
Jev is the judgment layer. Application code owns the allowed operations, and socai owns browser execution.
-
01
State the goal
Ask a concrete research question and, when useful, give a target count or stopping condition.
-
02
Jev chooses
Jev selects the platform, then one supported operation and one already observed target.
-
03
Chrome observes
The local socai CLI searches, opens, scrolls, and extracts through your existing browser session.
-
04
Evidence returns
Cards stream into the UI, the action set changes, and the final report cites captured source URLs.
THREE WORKFLOWS
One loop.
Platform-specific actions.
Only operations exposed by the installed socai CLI enter Jev's choice set.
Videos, authors, comments, media
Search and inspect observed videos or authors. Media download appears only when the original goal explicitly requests it.
- Starts with
- Search
- Media
- Explicit intent
People, content, companies
Choose the matching search type, then inspect an observed profile, company, post, experience, or education record.
- Search types
- 3
- Targets
- Observed URLs
Search cards are leads, not proof that a full post or comment thread was read. Jev Social labels partial output and keeps access gates, empty searches, interruptions, and step limits visible.
QUESTIONS THAT FIT
Ask for a finding you can check.
Good goals name the platform or let Jev route one clear platform, request a small evidence set, and say what should be compared.
- Use a concrete topic instead of a generic “research everything” request.
- Ask Jev to open relevant details and comments, not just collect search cards.
- Keep the first run small; four useful records are enough to inspect the workflow.
Instagram
Find four emerging AI creators. Open the most relevant posts and compare how people respond in comments.
TikTok
Find AI wearable videos. Inspect four results and summarize repeated claims, questions, and objections.
LinkedIn
Find AI product managers in San Francisco and compare the public experience and posts captured for each profile.THE BOUNDARY
Useful because the action space stays small.
The browser result is untrusted evidence. It can change what Jev sees next, but it cannot become an instruction.
-
01
No invented tools
Malformed, unknown, and low-confidence decisions stop before a CLI command runs.
-
02
No arbitrary targets
Action targets come from captured results or an explicit HTTPS social URL in the goal.
-
03
No account changes
The application does not post, like, follow, message, purchase, or change settings.
-
04
No fake success
Login gates, access limits, empty results, and interruptions return partial evidence.
TRY IT LOCALLY
Start with four records.
Use Node.js 22+, a current socai CLI, and Chrome already signed in to the platform you want to research. Onboarding configures the Jev decision provider and checks local readiness.
TERMINAL
npx github:socai-io/jev-social#v0.1.13 onboard
npx github:socai-io/jev-social#v0.1.13
npx github:socai-io/jev-social#v0.1.13 search "find emerging design creators on Instagram" --platform auto --limit 4