peerd – Browser-Native AI Agent Harness
peerd runs a full AI agent loop inside your browser as a Chrome/Firefox extension. No backend, no telemetry, BYOK to any model provider. Apache 2.0.
Tag
107 posts tagged #ai-agent
Browse 107 posts tagged AI Agent, including practical setup notes, reviews, comparisons, and workflow patterns for engineers working with AI tools.
peerd runs a full AI agent loop inside your browser as a Chrome/Firefox extension. No backend, no telemetry, BYOK to any model provider. Apache 2.0.
Sonarly connects to Sentry, Datadog, and Slack to triage alerts, identify root causes, and trigger AI fixes — cutting a 180-alert daily firehose down to 5 actionable issues.
Run AI agents in isolated, resumable MicroVM sandboxes with persistent filesystems, live run events, and a CLI that works with Claude Code, Cursor, and Codex.
BrowserOS is an open-source Chromium fork with a built-in AI agent. BrowserClaw gives your AI agents control of a real browser signed into your accounts. AGPL-3.0, 12K+ stars.
Runtime uses predefined DOM attributes as skill gateways to automate browsers, sidestepping expensive LLM reasoning at execution time.
Agnost AI continuously analyzes production conversations, catches agent failures your evals miss, and turns the highest-impact patterns into reviewed fixes.
Self-host a persistent AI agent with sandboxed execution, multi-channel messaging, voice, memory, and MCP tools — no Node.js required.
AI Agent Tools maps 510+ tools across 31 categories including MCP servers, frameworks, coding agents and platforms - updated daily. Find the right tool for any AI agent build.
An MCP server that gives AI agents full control of a real iPhone through macOS iPhone Mirroring. Screenshots, taps, swipes, and app control.
ScopeGate is an open-source MCP proxy that gives each AI agent exactly the access it needs — nothing more. Connect Google services, define granular OAuth scopes, and get an MCP endpoint URL in minutes.
Tambo is an open-source React toolkit that lets AI agents render dynamic UI components — charts, forms, maps — streamed as props in real time. 11K GitHub stars, MIT licensed.
Build MCP servers and apps in TypeScript or Python, debug in a browser inspector, and deploy to production with one CLI command via Manufact MCP Cloud. 10k+ GitHub stars.
An MCP server that gives AI agents real-time football scores, player lookups, and match summaries from Vublox. No API key needed.
Spectrum is an open-source multi-channel agent framework that routes AI agents through iMessage, Telegram, WhatsApp, Slack, Discord and other messaging platforms—no web chat required.
Connect any database to AI agents via auto-generated MCP or REST APIs. Supports PostgreSQL, MySQL, ClickHouse, Snowflake, BigQuery, and more with built-in PII protection.
Intuned uses an AI agent to generate production-ready Playwright scrapers in TypeScript or Python, with self-healing, captcha solving, and scheduling built in.
Papr gives AI agents persistent memory, knowledge connections, and context intelligence. Ranked #1 on Stanford's STaRK benchmark at 91% accuracy.
Feedster is an open-source RSS/feed reader with AI agent integration, sharable channels, and optional self-hosting via Docker.
Cua (YC X25) packages the sandbox, SDK, and benchmarks you need to build, evaluate, and deploy AI agents that control full desktops — macOS, Linux, Windows.
One binary Rust agent server with sandboxed execution, multi-provider LLM support, voice, memory, MCP tools, and built-in integrations for Telegram, Discord, and more.
owthorize is a synchronous Node.js guard that parses SQL, HTTP, shell, and filesystem tool calls into typed shapes and evaluates them against rules — catching destructive AI agent actions before they execute.
peerd is a browser extension that turns Chrome or Firefox into a full AI agent harness with sandboxed compute, WebRTC P2P sharing, and no backend.
ClawNet gives AI agents their own @clwnt.com email address, DMs, and calendar. Plain HTTP with bearer tokens — no OAuth, no SDK, no WebSocket.
Nelly is a cross-platform AI agent builder that lets you create, test, and deploy teams of agents using natural language — no coding required. Currently in beta for macOS, Windows, and Linux.
Chamber is an AIOps platform that puts an AI agent between your team and your GPU fleet. Ask in plain language, get root-cause analysis and automated remediation across Kubernetes and Slurm.
Record browser workflows or describe them in plain English. Tasker is a free, MIT-licensed desktop agent that uses AI to execute multi-step web tasks. Supports Claude, GPT-4o, and Gemini.
AgentLair gives AI agents a persistent email identity, encrypted credential storage, and namespace isolation via a single API — no OAuth, no human in the loop.
See cloud cost estimates for Terraform before deploying. Infracost plugs into Claude Code, Cursor, CI/CD pipelines, and your terminal.
Open-source MCP server that scores AI agent output quality, catches safety failures like PII leaks, and enforces cost budgets across traces.
Generate MCP-compatible tools from any OpenAPI spec and connect to Claude, Cursor, or any MCP client. Official Taskade MCP server with 62 built-in tools.
A Chrome Extension that exposes your browser as a remote MCP server, letting any AI agent drive web tasks without context switching. BYO-subscription model.
Voygr gives AI agents and apps fresh, queryable place profiles combining accurate location data with news, events, and web context that Google Maps APIs cannot.
ROS-MCP-Server bridges Model Context Protocol to ROS, letting Claude, GPT, or Gemini control physical robots without touching robot source code. Setup uses rosbridge, works across ROS 1 and ROS 2.
DeepReel's AI Video Genie transforms blog posts, PDFs, and plain text into polished explainer videos with AI presenters, voiceovers, captions, and B-roll — in minutes.
Build and run AI agentic workflows across channels with YAML definitions, MCP integration, memory support, and multi-channel continuity in one Node.js runtime.
Point WebMCP Core at any URL, get AI agent tool definitions in seconds. Auto-crawls forms, APIs, and interactive elements; outputs TypeScript, React hooks, or YAML for Chrome 146+.
Reverse-engineer any platforms internal APIs by recording browser traffic with Integuru, an open-source AI agent.
Spine Swarm runs parallel AI agents across 300+ models on an infinite visual canvas, turning complex non-coding projects into composable block-based workflows.
Intuned is a code-first browser automation platform with an AI agent that generates Playwright scrapers, crawlers, and RPA bots from natural language prompts—and keeps them running when sites change.
An AI browser agent that browses the web in plain English. Telegram bot, REST API, free tier. Automates scraping, bookings, forms, and price monitoring without code.
OpenMolt is a Node.js framework for building AI agents with a self-directed reasoning loop, 30+ built-in integrations, and scope-gated security.
Production-grade middleware that intercepts AI agent tool calls, evaluates them against policy, and blocks or logs risky actions before execution.
Gabriel Operator is a teach-by-demo browser automation tool. Click through a task once and it becomes a reusable AI-powered workflow you can run with parameters, loops, and branches.
DroidClaw uses ADB and a perception-reasoning-action loop to let an LLM control your Android phone by tapping, typing, and swiping. No API integrations needed.
Skyvern automates browser-based workflows using AI that sees pages like a human, adapting to layout changes without brittle selectors. Open source with 21.9k GitHub stars.
Describe what you need. Intuned Agent generates production-ready Playwright code, deploys it, and fixes it when sites change. Built-in stealth, auth, scheduling, and scale.
Sonarly (YC W26) connects to Sentry and Datadog to triage production alerts, deduplicate noise, and auto-fix bugs with Claude Code. Cut MTTR with AI root cause analysis.
Open-source Agent Alignment Protocol (AAP) makes AI agent behavior observable with Ed25519-signed decision traces, EU AI Act compliance, and cross-provider verification.
boo is a GNU screen-style terminal multiplexer written in Zig on top of libghostty-vt, with a faithful terminal state model and agent-friendly peek, send, wait, and JSON primitives.
WeSight is an open-source Electron desktop console for Claude Code, Codex, Kimi Code, OpenClaw, Hermes Agent, OpenCode, and other local AI coding agent runtimes.
Skyvern is an open-source AI agent that automates complex browser-based workflows using LLMs, supporting multi-step tasks, form filling, and API integrations.
DeepReel turns blog posts, prompts, and files into professional AI videos with avatars, voiceovers, and visuals. No editing skills needed.
Flowtest.ai is an AI agent that browses your website like a real user, detects issues, and self-heals when layouts change.
Scalable browser infrastructure for AI agents. Launch headless Chrome sessions via API, run web automation at scale without managinginfra.
A code-first browser automation platform with an AI agent that builds and maintains your automations as deterministic, production-ready code.
An MCP server that adds a cost signal to AI agents - they plan upfront, stay in budget, and stop when done. Works with any MCP client.
Capframe maps every tool your AI agents reach, mints scoped capability tokens in Rust, and enforces runtime policy decisions in microseconds—no LLM in the path.
Familiar is an open-source local AI agent that runs entirely on your Mac or iOS device — no cloud, no data leaving your hardware. Build and orchestrate agent workflows privately.
Nanobot bridges MCP servers and AI agents, letting you turn any MCP-compatible tool into an autonomous agent with memory, planning, and tool use.
A cloud-hosted AI agent that builds, deploys, and manages SaaS tools with minimal manual ops. Useful for indie hackers who want an AI teammate across build and launch workflows.
The most comprehensive map of the AI agent ecosystem — 510+ tools across 31 categories including MCP servers, coding agents, and frameworks. Updated daily.
Port42 watches your actual work patterns and automatically generates tools agents can use. No manual tool registration, no API wrappers—just let it learn how your team operates.
Intuned Agent converts natural language prompts into production-ready Playwright code, fixes flows when sites change, and deploys with built-in stealth, scheduling, and auto-scaling.
Zatanna reverse-engineers any web portal's network traffic and generates headless APIs that AI agents can execute in seconds. No UI required.
Comprehensive directory of 510+ AI agent tools spanning reasoning, tool use, memory, and multi-agent frameworks — organized by category with direct links to docs and GitHub.
MCP server that pauses AI coding agents when stuck, creates markdown Q&A for human clarification, then resumes. Zero-config, cross-platform, 150 GitHub stars.
YC S22 startup Intuned lets you build, deploy, and maintain browser automations as code. An AI agent writes Playwright scripts and auto-heals them when sites change.
Nimbus is a desktop browser that replaces the URL bar with an AI chat bar, letting an agent navigate, fill forms, and extract data. A UX experiment inspired by Claude Code.
Augento fine-tunes open-source LLMs for AI agents using reinforcement learning. Drop in a reward function, no dataset required. YC W25, 101pts on Launch HN.
Jido is a 1.7K-star Elixir agent framework on the BEAM. Pure-functional agents, GenServer supervision, MCP, durable workflows, and 1,722 GitHub stars.
Trigger.dev builds and runs reliable AI agents and long-running workflows in TypeScript. Apache 2.0, no timeouts, 15K stars, MCP server, self-host or cloud.
Relvy is an AI agent that automates on-call runbooks, analyzing telemetry data and code at scale to help DevOps teams resolve incidents faster.
Freestyle gives coding agents their own Linux VMs with instant horizontal forking — fork mid-browser-session, snapshot any state, provision in 500ms. Built on self-hosted bare metal.
Port42 is a native Mac app where AI companions build interactive UIs and act on your machine through a single API. They ship 40+ releases in 9 days alongside you.
Leap is an AI developer agent that builds and deploys real backend apps to your own AWS or GCP cloud. Connect GitHub, get PR-style diffs, isolated previews, and architecture diagrams.
Second connects your GitHub account to AI bots that generate new web applications or add features to existing ones by creating and modifying code files, then raising PRs directly.
Moltis is a Rust-native AI agent server with memory, sandboxed execution, multi-provider LLMs, and MCP tool servers. One binary, no runtime deps.
Browse, discover, and safely run AI tools with cryptographic verification. Enact is the npm for AI agents—a registry with end-to-end proofs that tools do.
Vela is a YC W26 AI agent that reads your emails, calendars, and communication channels to automatically book meetings across candidates, clients, and time.
Propolis runs swarms of browser agents that collaboratively explore your web app, report bugs, and write end-to-end tests for your CI pipeline.
An AI data analysis agent for business users. Upload CSV, connect SQL databases, and describe what you want to find. Charts, stats, and reports generated.
World2Agent standardizes how AI agents perceive the real world — screenshots, device state, sensor data — into a typed protocol. TypeScript SDK with React.
The official Python SDK for Google Antigravity. Build AI agents with Gemini, MCP tools, custom skills, and managed infrastructure. Async-first API with typed.
Harmonist orchestrates up to 186 AI agents with mechanical protocol enforcement and dependency management. Define agent workflows in YAML — portable across.
Future AGI is a self-hostable platform for evaluating, observing, and improving LLM apps. Includes tracing, evals, simulation, datasets, an AI gateway, and.
DeepSeek-Reasonix is a TypeScript agent framework built for DeepSeek models with structured reasoning, tool calling, and multi-step planning.
A public knowledge base of 150+ agent design patterns from dozens of production deployments. Covers planning loops, tool design, prompt engineering.
Terax AI is a 7MB terminal-first AI dev workspace built with Tauri and Rust. Bundles AI agents, code editing, and terminal into one cross-platform desktop app.
Mercury Agent runs 24/7 from CLI or Telegram with a personality system, token budgets, and permission-hardened tools. Built with TypeScript and Docker with BYO.
Garden Skills bundles four Agent Skills for Claude Code, Cursor, and Codex: web design, video presentations, GPT Image 2 prompting, and local knowledge.
Nomi joins your sales calls and gives real-time phrase suggestions when prospects push back, signal interest, or are ready to close.
TeamOut is an AI agent that plans company retreats entirely through conversation — handling venue sourcing, vendor coordination, flight cost estimation, and.
Crustdata maps web search results to real-world entities (people, companies, events) so AI agents spend tokens on reasoning, not on cleaning up messy HTML and.
Runtime orchestrates coding agents across sandboxes like E2B and Daytona, handling environment setup, secrets injection, and RBAC so your whole team ships with.
Embedder is a hardware-aware AI agent that reads datasheets, writes firmware code, flashes the board, runs tests, and autonomously fixes bugs across 500+ MCUs.
Promptless watches your PRs and support tickets, then automatically drafts and pushes documentation updates to Readme, Mintlify, Intercom, and more.
Terminal Use lets you deploy AI agents that read and write files — coding agents, research agents, and internal tools — in sandboxed environments.
Runtime lets engineering teams deploy sandboxed coding agents (Claude Code, Codex) where non-engineers can safely ship code without risking production or.
Caddie is a Slack-native AI agent that connects to90+ tools via MCP and automates GTM tasks like CRM logging, cold outreach, and meeting prep with plain.
Codebuff brings Cursor Composer-style AI code generation to your terminal. Describe what you want in plain English and it edits your codebase directly — no.
Vela is a YC W26 AI agent that handles multi-party, multi-channel scheduling across email, SMS, WhatsApp, Slack and phone — booking interviews, following up on.
AgentMail gives AI agents their own email inboxes via API. Create domains, configure webhooks, send and receive emails programmatically. From YC S25.
HumanLayer is a YC-backed API that lets AI agents request human approval at critical steps during execution, enabling safe production deployment of autonomous.
CloudRouter is a skill and CLI that lets AI coding agents like Claude Code spin up cloud VMs with GPUs on demand, run dev servers and browser automation, then.
Modulus is a desktop app that connects multiple AI coding agents (Claude Code, Cursor, Devin) to shared project memory, solving cross-repository context loss.
Voygr is a YC W26 company building a high-fidelity location data layer for AI agents and apps. It validates and enriches place records — addresses, hours.
Sonarly triages alerts, finds root causes, and opens fix PRs on GitHub. 40+ integrations, 84% root-cause accuracy, cuts MTTR 10x. YC W26.