Agent apps.
One command away.
Experiences built for agents, not browsers. Every app installs as a sha256-pinned, signature-verified native service — auto-spawned by the daemon, callable over typed IPC in seconds.
An agent app is an installable capability — a database, a search backend, a sandbox, a payment rail — packaged as a typed JSON-in/JSON-out service that runs locally on your daemon. The loop is discover → install → call: browse the catalogue, install with one command, and call the app's methods from any agent on the network. New to the model? Read what an agent app store is or jump to turning your own API into an agent app.
PostgreSQL
Run and query PostgreSQL from an agent — local server lifecycle + psql, any libpq target.
New & Updated
Just shipped or recently bumped on the catalogue.Data & Storage
Databases, caches, and stores an agent can stand up in one command.AI & Search
Grounded search, retrieval, and generation for agents.Web & Automation
Turn the live web into clean text and drive real browsers.Infrastructure
Containers, microVMs, and deploys — the compute layer for agents.Security
Guardrails and firewalls for agents and their inputs.Finance & Payments
Settle value on-overlay and read the markets.Work & Research
Put real-world work in motion and get real answers back — human hands on a task, and pricing research with your own customers.Communications
Give an agent its own phone number or email inbox — voice, SMS/iMessage, email, and threaded conversations.All apps
Ship an agent app.
It’s just a manifest.
Describe your app's methods in a guided submission — we build, sign, and review the adapter. Once approved, pilotctl appstore install <id> works everywhere on the overlay. Wondering how agent apps fit next to MCP tools? See MCP + Pilot: tools and a network for AI agents.
