NotASithLordBrowser & desktop

peerd

A browser-native AI agent runtime that drives tabs, executes sandboxed compute, and manages tasks directly within your browser via an extension.

Frontend
DevOps
Automation
AI coding
Security
Browser extension
Browser
Runs locally
Needs API key

Open website View on GitHub

Popularity: 401 Stars; GitHub stars
Recent activity: 9/1/2026; Updated in the last 30 days
License: APACHE-2.0; Permissive

Why it matters

We look beyond stars: what problem it solves, whether it creates real utility, and what makes its approach worth noticing.

Last 90 days

Problem

Traditional AI agents often require independent cloud environments or complex desktop permissions, making it difficult to safely and directly operate everyday browser sessions.

Practical value

Embeds the agent runtime directly into a browser extension, leveraging existing tabs, sandboxed compute, and BYOK model connections for practical in-browser automation.

Innovation / differentiation

Pulls the agent architecture into native browser primitives like Workers, OPFS, and WASM, achieving in-browser distributed collaboration without a central backend or telemetry.

Leverage potential

Supports Chrome and Firefox with a modular extension architecture, enabling developers to hook into low-level agent loops, sandbox tools, and peer-to-peer networking.

Why now

With advancing browser-based model inference, WebGPU, and WASI standards, the timing is ripe for running agents directly inside the browser.

Community activity

In the last 90 days there were 98 new issues and 362 pull requests; the bounded issue/PR samples include 5 issue authors and 4 PR contributors, with 22 releases.

Maintainer responsiveness

The 98-issue window sample had a 73% close rate, and the 100-pull-request sample had a 87% merge rate. Maintainer-response observations covered 65% of that issue sample, with a 19% response rate and median first response of 3.8 hours.

4 contributors in PR sample22 releasesIssue response rate 19% · sample 64 (65% coverage)Median first response 3.8hPR merge rate 87% · 100 sampled in window

Key highlights

Drives browser tabs and signed-in sessions directly
Built-in sandboxed compute (JS Notebooks, WASM Linux VMs)
Agent execution isolated via browser primitives

Quick start

How it is installed, how hard it is, and where to start.

Where it runs

Browser

Difficulty

Medium — some setup needed

Runs locally

01Download the source code or release package from GitHub and load it via your browser's extension management page.
02Open the extension and create a local vault with a secure passphrase.
03Configure your model provider API key or select a local model in the settings.
04Launch the agent from the browser sidebar and begin your tasks.

Best for

Developers who want to try it on their machine

More about it

peerd is an AI agent runtime that lives inside your browser, pulling the agent harness into the browser rather than connecting the browser to an external platform. By utilizing native browser isolation boundaries like Workers, OPFS, and WASM, it securely drives web pages, executes code, and manages complex tasks without exposing user credentials.

The project runs entirely locally with no backend dependency and zero telemetry. Users can choose between cloud-based model APIs or local models, with all sessions, memory, and task states managed within the extension's encrypted local vault.

Sources

Each field shows its status and source — expand to review.

14 · Expand

capability tags
Verified
automation, ai_coding, security
Source: admin_cms · cms editor · 8/27/2026
editor note
Verified
{"en":"This project leverages browser-native primitives like Workers and WASM for agent isolation instead of relying on external servers. It is ideal for users requiring deep web interaction with strict data privacy.","zh":"该项目通过浏览器原生机制（如 Workers 和 WASM）实现 Agent 隔离，而非依赖外部服务器。适合需要深度网页交互且对数据隐私有严格要求的用户。"}
Source: admin_cms · cms editor · 8/27/2026
how to use
Verified
{"steps":[{"en":"Download the source code or release package from GitHub and load it via your browser's extension management page.","zh":"从 GitHub 下载源码或发布包，并在浏览器扩展管理页面加载。"},{"en":"Open the extension and create a local vault with a secure passphrase.","zh":"打开扩展并创建本地保险库，设置访问密码。"},{"en":"Configure your model provider API key or select a local model in the settings.","zh":"在设置中配置模型提供商 API Key 或选择本地模型。"},{"en":"Launch the agent from the browser sidebar and begin your tasks.","zh":"在浏览器侧边栏启动 Agent 并开始交互。"}],"installAt":"browser","difficulty":"medium"}
Source: admin_cms · cms editor · 8/27/2026
intro
Verified
{"en":"peerd is an AI agent runtime that lives inside your browser, pulling the agent harness into the browser rather than connecting the browser to an external platform. By utilizing native browser isolation boundaries like Workers, OPFS, and WASM, it securely drives web pages, executes code, and manages complex tasks without exposing user credentials.\n\nThe project runs entirely locally with no backend dependency and zero telemetry. Users can choose between cloud-based model APIs or local models, with all sessions, memory, and task states managed within the extension's encrypted local vault.","zh":"peerd 是一款运行在浏览器内部的 AI Agent 运行时，它将 Agent 的执行环境直接植入浏览器，而非将浏览器接入外部平台。通过利用浏览器原生的隔离边界（如 Workers、OPFS 和 WASM），它能够在不泄露用户凭据的情况下，安全地驱动网页、执行代码并管理复杂的自动化任务。\n\n该项目完全在本地运行，不依赖后端服务器，也不收集任何遥测数据。用户可以根据需求选择云端模型 API 或本地模型，所有会话、记忆和任务状态均存储在浏览器扩展的加密保险库中。"}
Source: admin_cms · cms editor · 8/27/2026
Latest release
Verified
v0.7.3
Source: GitHub API · latest_release=v0.7.3 · 9/1/2026
License
Verified
Apache-2.0
Source: GitHub API · license.spdx_id=Apache-2.0 · 9/1/2026
needs api key
Verified
Yes
Source: admin_cms · cms editor · 8/27/2026
One-liner
Verified
{"en":"A browser-native AI agent runtime that drives tabs, executes sandboxed compute, and manages tasks directly within your browser via an extension.","zh":"一款浏览器原生 AI Agent 运行时，通过扩展程序直接在浏览器内驱动标签页、运行沙箱计算并管理任务。"}
Source: admin_cms · cms editor · 8/27/2026
Platforms
Verified
browser
Source: admin_cms · cms editor · 8/27/2026
Category hint
Inferred from materials
ai-apps
Source: Project README · hint=ai-apps · 8/15/2026
product forms
Verified
browser_extension
Source: admin_cms · cms editor · 8/27/2026
role tags
Verified
frontend, devops
Source: admin_cms · cms editor · 8/27/2026
supports local
Verified
Yes
Source: admin_cms · cms editor · 8/27/2026
supports self host
Verified
No
Source: admin_cms · cms editor · 8/27/2026

Other verified projects matched by category, capabilities, and intended roles.

SwiftBar

An open-source macOS utility that turns scripts into custom menu bar items and supports the BitBar/xbar plugin format.

Same Browser & desktop categoryBoth support Automation

PowerToys

A Microsoft-maintained set of Windows utilities for window layouts, keyboard shortcuts, file tasks, and everyday system workflows.

Same Browser & desktop categoryActive in the last 30 days

Linguist

An open-source browser translation extension for pages, selections, inputs, and subtitles, with on-device offline translation and custom backends.

Same Browser & desktop categoryActive in the last 30 days

Flare

A graph-first IDE for agentic coding featuring a live codebase map, integrated terminal, and MCP task collaboration.

Both support AutomationBoth fit Frontend

OpenSandbox

An open-source sandbox platform for AI agents and code execution with Docker/Kubernetes runtimes, SDKs, CLI, and MCP integration.

Both support AutomationBoth fit DevOps

omnigent

An open-source orchestration layer for running, switching, and collaborating across multiple AI agents through terminal, web, desktop, and cloud-sandbox sessions.

Both support AI codingBoth fit DevOps