jelly
Browser instrumentation for agents.
Install / Use
claude mcp add FabioFlorey -- npx -y github:FabioFlorey/jellyIf the server publishes to npm under a different name, use that package instead — check the repo README.
MCP Server
Model Context Protocol server
Quality Score
Category
AutomationSupported Platforms
Tags
Our assessment of jelly
jelly scores 72/100 on our quality scale, 2631st of 2,887 Automation skills we index.
Its MCP Server is 6.1 KB long, split into 5 sections with 6 code examples: a thorough specification that gives an agent plenty to work with.
It has 3 GitHub stars, so there is little community track record yet; judge it on its content.
Maintenance, license and trust
- The repository was last updated 4 days ago, so jelly is actively maintained.
- No license is declared. By default that means all rights are reserved: you can read it, but reusing or redistributing it is not clearly permitted. Ask the author before building on it commercially.
- Its trust signals score 75/100, with 3 cautions from licensing, adoption, age or documentation. These come from repository metadata, not a code audit — read the skill file before letting an agent act on it.
jelly compared with similar skills
All 4 of these similar skills score higher than jelly; compare them before choosing.
| Skill | Score | Stars | Updated | Format |
|---|---|---|---|---|
| jelly (this skill)by FabioFlorey | 72 | 3 | 4d ago | MCP Server |
| Agent-Reachby Panniantong | 100 | 93.0k | 21d ago | CLAUDE.md |
| headroomby headroomlabs-ai | 100 | 74.6k | today | CLAUDE.md |
| CowAgentby zhayujie | 100 | 47.3k | today | CLAUDE.md |
| Scraplingby D4Vinci | 100 | 86.1k | today | MCP Server |
Frequently asked questions
- How do I install jelly?
- Run
claude mcp add FabioFlorey -- npx -y github:FabioFlorey/jelly. The install tabs above show the steps for each supported agent. - Which AI agents does jelly work with?
- It is written for Claude Code and Claude Desktop, as a MCP Server file. Other agents that read the same format can often use it too.
- Is jelly safe to use?
- It declares no license and scores 75/100 on trust signals. Skills are instructions an agent will follow, so read the file before installing it and do not approve commands you do not understand.
- Is jelly still maintained?
- The repository was last updated 4 days ago, so jelly is actively maintained.
Skill content
View source on GitHubBrowser instrumentation for agents.
✨ Quickstart • ⭐ Architecture • ⚡ MCP • 💡 Documentation • 🍯 Tool Index
</div>1. About
Jelly is a small Rust toolkit that gives AI agents a practical way to observe and control a real Chromium browser.
It turns browser operations such as reading pages, finding interactive elements, clicking, typing, navigating, switching tabs, uploading files, inspecting network traffic, and taking screenshots into reusable primitives that agents can discover and compose.
You do not need to know Chrome DevTools Protocol to use it. Jelly's MCP server now defaults to the small-surface Agent API: browser-schema discovers semantic operations, browser-call executes ordered semantic batches, and browser-events handles retained CDP notifications. The expanded individual-tool API is available as large-surface, and raw CDP stays separately opt-in.
<p align="right"><sub><a href="#top">⭐ Back to top</a></sub></p>[!CAUTION] Jelly controls a real browser and can perform real actions on websites. Review routines before running them against accounts or systems you care about.
2. Quickstart
Interactive setup:
git clone https://github.com/FabioFlorey/jelly.git
cd jelly
./quickstart.sh
The wizard checks dependencies, configures OAuth + hosting, and can install/start Jelly. See MCP for hosting modes and advanced setup.
Use ./quickstart.sh --dry-run to preview setup, or ./quickstart.sh --check for prerequisites only.
Or build the tools directly:
cargo build --bins
Open a browser and inspect a page:
cargo run --quiet --bin agent-run -- \
open-browser https://example.com
cargo run --quiet --bin agent-run -- read-page
cargo run --quiet --bin agent-run -- snapshot-interactive
Act on a discovered element using the ref returned by snapshot-interactive:
cargo run --quiet --bin agent-run -- click '<ref>'
Runtime refs are document-scoped tokens such as @eabc123-7; legacy rollback mode uses numeric refs such as @e7.
Close the browser:
cargo run --quiet --bin agent-run -- close-browser
Discover tools without loading the full catalog:
cargo run --quiet --bin agent-discover -- capabilities
cargo run --quiet --bin agent-discover -- search "find form inputs"
cargo run --quiet --bin agent-discover -- schema type-text
<p align="right"><sub><a href="#top">⭐ Back to top</a></sub></p>
3. Documentation
<div align="center">| Document | Description | | :--- | :--- | | 📒 Documentation Index | Main entry point for the documentation | | ⚠️ Requirements | What must be installed before running Jelly | | 💡 Glossary | Definitions for technical and project-specific terminology | | ⭐ Architecture | Browser session, primitives, registry, and execution model | | 🔑 Tool Discovery | Capability discovery, search, and schema loading | | 🔆 Routines | Guarded workflow graphs, branching, loops, and HITL continuation | | 🍯 Reliability | Verification, typed failures, artifacts, timeouts, and cleanup | | ⚙️ Configuration | Versioned technical configuration, secrets/deployment environment, and runtime roots | | 📂 Runtime Layout | Runtime state, build output, logs, screenshots, and cleanup | | 🧈 Development | Project conventions, tests, and contribution rules | | 🧪 Test Index | Test-suite layout, coverage groups, batches, and executable catalog | | ⚡ MCP Server | Small-surface Agent API, large-surface mode, authentication, tool mapping, and deployment boundary | | 🍯 Tool Index | Generated reference for browser primitives and system tools | | 🌟 Changelog | Development history |
</div> <p align="right"><sub><a href="#top">⭐ Back to top</a></sub></p>4. Live demo
<div align="center"> <a href="https://fabioflorey.github.io/jelly/"> <img src="./assets/jelly-demo.gif" alt="Jelly controlling a real Chromium browser" width="880"> </a><sub><strong>Jelly documenting Jelly.</strong> This video was generated by Jelly itself from a real Chromium session. After each browser action, Jelly waits for the page to settle, captures the resulting state, labels what happened, and assembles the steps into this self-documenting trace.</sub>
</div> <p align="right"><sub><a href="#top">⭐ Back to top</a></sub></p>5. Status
Jelly is under active development. Interfaces may change while the execution, discovery, and routine layers settle.
Jelly is currently distributed under a proprietary All Rights Reserved license. See LICENSE for the applicable terms.
Project contact: jelly@fabioflorey.com.
<p align="right"><sub><a href="#top">⭐ Back to top</a></sub></p>Related Skills
Agent-Reach
93.0kGive your AI agent eyes to see the entire internet. Read & search Twitter, Reddit, YouTube, GitHub, Bilibili, XiaoHongShu — one CLI, zero API fees.
headroom
74.6kCompress tool outputs, logs, files, and RAG chunks before they reach the LLM. 20% fewer tokens for coding agents, 60-95% fewer tokens for JSON, same answers. Library, proxy, MCP server.
CowAgent
47.3kOpen-source personal AI assistant & Agent Harness. Plans tasks, runs tools and skills, self-evolves with memory and knowledge. Multi-agent, multi-model, multi-channel. Lightweight, extensible, one-line install.
Scrapling
86.1k🕷️ An adaptive Web Scraping framework that handles everything from a single request to a full-scale crawl! Don't be shy, join here: https://discord.gg/EMgGbDceNQ and follow here for daily tips and tricks: https://x.com/Scrapling_dev
Languages
Trust signals
From repository metadata: license, adoption, age and documentation. Not a code audit — see the Safety scan above for what the skill file itself contains.
