by andududu
An unofficial local visualizer for Jev (TypeSafe): a live view of every call your code makes. Not affiliated with TypeSafe AI.
# Add to your Claude Code skills
git clone https://github.com/andududu/jeviewSee how jeview compares with popular alternatives.
jeview is an open-source ai agents skill for AI coding assistants such as Claude Code, Codex CLI, and ChatGPT, built by andududu. An unofficial local visualizer for Jev (TypeSafe): a live view of every call your code makes. Not affiliated with TypeSafe AI. It has 50 GitHub stars.
jeview's catalog security scan is still queued. You can run an instant dependency and prompt-injection check now with the "Scan for vulnerabilities" button above.
Clone the repository with "git clone https://github.com/andududu/jeview" and add it to your Claude Code skills directory (see the Installation section above).
jeview is primarily written in JavaScript. It is open-source under andududu on GitHub, so you can review or fork the full source.
Yes. SkillsLLM lists many other AI Agents skills you can browse and compare side by side. Open the AI Agents category from the badge at the top of this page, or use the Related Skills and comparison links further down to weigh jeview against similar tools.
No comments yet. Be the first to share your thoughts!
⚠️ Third-Party Software Notice
This skill is third-party open-source software developed and hosted independently on GitHub. SkillsLLM is an informational directory and does not control or maintain the underlying repository.
Any security checks, ratings, or warnings displayed by SkillsLLM are automated and limited in scope. They do not constitute a security certification or guarantee that the software is safe, error-free, or free from malicious code, vulnerabilities, compromised dependencies, or prompt-injection risks.
Review the source code, permissions, dependencies, and configuration before installing or running any third-party skill. Use is at your own risk. To the maximum extent permitted by applicable law, SkillsLLM is not liable for losses arising from third-party software.
The deep catalog scan for this skill is still queued. Run an instant dependency check now instead.
Agents: llms.txt says how to use a running Jeview, and AGENTS.md how to work on this repository.
An unofficial local visualizer for Jev (TypeSafe System One): a live view of every call your code makes. Jeview is an independent project. It is not affiliated with TypeSafe AI, nor made or endorsed by it.
your code → Jeview (127.0.0.1:4777) → TypeSafe (api.typesafe.ai)
Jeview is a gateway: it sits between your code and TypeSafe and answers nothing itself. Point your Jev client at Jeview instead of TypeSafe. Jeview sends each request on to Jev with your key, hands Jev's answer back, keeps every call in a local SQLite database, and draws the calls on a live map as they happen.

Needs Node 24 or later, and nothing else: Jeview has no dependencies to install.
./launch.sh
That finds a suitable Node (on your PATH, or installed by nvm or Homebrew), starts Jeview and opens the viewer at
http://127.0.0.1:4777/ (or http://jeview.localhost:4777/, a name your browser already knows). npm start does the same
without opening a browser. Add your TypeSafe API key (the key icon,
top right). The first time you open it, the viewer explains itself.
Two demos send real Jev calls through Jeview. Run one in a second terminal and put its window beside Jeview's. Each plays only while its page is open, costs about ten cents an hour while it does, and stops with Ctrl-C.
npm run demo, then open http://127.0.0.1:4781/: Jev plays Pixel Knight, a tiny side-scroller, deciding every move.npm run demo:support, then open http://127.0.0.1:4782/: Jev triages a made-up support inbox, deciding what each
message is about, how urgent it is, which team takes it and whether to offer a credit.Send Jev requests to http://127.0.0.1:4777/v1/systemone instead of https://api.typesafe.ai/v1/systemone: the same
requests and the same answers, with no key needed from the caller.
http://127.0.0.1:4777/my-project/v1/systemone.
The viewer can then show one group at a time.events. When a later request follows from one of
those answers, send its event id in a Jeview-Trigger header, and the map grows that request off the answer.
Jeview drops its own headers before calling Jev and sends the body on unchanged.c14. When the criteria behind
the keys are objects, a Jeview-Display header says which part to show instead: Jeview-Display: name, or a field
per question, Jeview-Display: category=name, kind=title. An option without that field keeps its key, and so does
everything sent without the header. The header is only for show: one that cannot be read is ignored, never refused.Agents can read http://127.0.0.1:4777/llms.txt: a running Jeview serves it with its own address and whether a key is
set. llms.txt here is the same text, for the default address.
Given to either launcher: ./launch.sh --port 4800, or npm start -- --port 4800.
--port 4777--dir ~/.local/share/jeview: where the database lives--jev-endpoint URL: another Jev endpoint, such as a local mockNothing is stored anywhere but your own machine. Jeview runs locally, has no accounts and no tracking, and the only place it sends anything is TypeSafe, to make the calls you asked for.
Calls and your key are kept in jeview.sqlite in the data folder. The file is readable only by you, and the key is
stored in it as plain text. To erase everything, key included, stop Jeview and delete jeview.sqlite* from that folder.
A long history stays cheap: the viewer opens on the latest 20,000 calls, and search reaches the rest. Two Jeviews may share a data folder, for instance on two ports, and each shows the calls of both.
Jeview listens on 127.0.0.1 only and has no login: anything running on your machine can read the recorded calls and send calls with your key. It refuses requests from web pages on other sites, so a page you visit cannot. Do not put it behind a public address.
npm install # only the type checker
npm test
npm run typecheck
npm run check:pages # the viewer's and the demos' browser scripts parse
MIT