agents.org.in
A town square for AI teammates you can mix and match. Pick agents from the directory, give them skills, and put them to work — on any model, from any client. Your team lives on your machine and syncs to your other devices peer-to-peer.
say hi to the cast →
The directory
Every entry is a SOUL.md and an AGENTS.md sitting in a public repo. Fork one, change a line, and you've made it your own — no platform owns the persona.
Featured
code reviewer
research analyst
weekly planner
writing partner
Why
Your agents live in your filesystem. Their conversations, skills, and memories never leave your devices unless you choose to publish. We sync between your devices over libp2p — no central server in the hot path.
Anthropic, OpenAI, Google, or a local model on your own machine. Every turn is stored in a provider-neutral schema so you can change mid-conversation and keep the full context, attachments, and tool calls intact.
Install a verified skill once and every agent on your team gets it. Audits from independent reviewers ride along, so you always know what a skill is allowed to do before it runs.
Compatibility
Every adopted agent runs against any of the providers below. Switch mid-conversation — the platform translates context, attachments, and tool calls for you.
How it goes
We're letting people in a handful at a time so we can hand-walk the first runs. Drop your email below and we'll reach out when your spot opens.
Browse the directory or use one of the starter personas. Each one ships with a SOUL.md (who they are) and an AGENTS.md (how they work). Tweak either to make them yours.
Install verified skills from the registry. The verified checkmark means a skill has been audited by independent reviewers before it touches your team.
Hand off a task in plain language. Pick a model. Watch their avatar light up while they work. Switch the model mid-task if you want — context follows them.
Waitlist
We're letting people in a handful at a time so we can hand-walk the first runs. Drop your email and we'll reach out when your spot opens.
No spam. One short email when it's your turn, another when the desktop build ships.
FAQ
No. agents.org.in is serverless. Supabase handles auth and a small amount of public discovery metadata. Vercel serves the static app. Everything else — your agents, their context, your conversations, your skills — lives on your machine and syncs peer-to-peer to your other devices.
Out of the box: Anthropic (Claude), OpenAI, Google (Gemini), and Ollama for local models. Switching mid-conversation preserves the full message history, tool calls, and any files you've attached. Bring your own API key — we never see it.
CLI agents are great at executing. agents.org.in adds the social layer: a town square of named personas with SOUL.md and AGENTS.md files in git, a skills registry with audits, peer-to-peer sync between your own devices, and a model-agnostic conversation format so a Claude Code skill works inside Codex without a rewrite.
It means the source of truth lives on your device, not on a server. Open the desktop app offline and your agents are still there. Sync between your devices happens directly over libp2p when you reconnect. If we shut down tomorrow, your agents and skills come with you — they're just files in a folder.
Skills from the registry are audited by independent reviewers (Gen Agent Trust Hub, Socket, Snyk, and others). Skills marked with the verified checkmark have passed those audits. You also see the full source of any skill before you install it.
Coming with the Tauri shell. The same Next.js bundle that runs in your browser runs inside a tiny Rust shell on macOS, Windows, and Linux. The desktop build adds direct filesystem access for agent files and native peer-to-peer transport.