A versioned library of the skills your agents compile
Every compiled skill is stored, versioned, and reusable — browse them, run them directly, and share them across your agents.
GET /api/v1/skillsWhat agent & skill library does
As your agents work, they build a library of compiled skills. Twin keeps that library organized: each skill is versioned, named, and runnable on its own, so a flow one agent figured out becomes a building block every agent can call.
Versioned skills
Each skill carries a stable id and version, so you can pin, roll forward, or audit exactly what ran.
Run a skill directly
Skills are first-class — invoke one by id with typed inputs from the API or over MCP, no goal needed.
Reusable across agents
A skill compiled by one agent is available to the rest, so your fleet stops re-solving the same flows.
Browse in the dashboard
See every skill, its versions, and recent runs in one place — with the session video for each.
From a goal to deterministic action
- 1CompileA successful run compiles into a named, versioned skill in your library.
- 2OrganizeSkills are stored per tenant with versions and run history.
- 3ReuseAny agent — or a direct API/MCP call — can run a skill by id with typed inputs.
- 4IterateReal drift bumps a version; you can pin to a known-good version anytime.
See it on a real call
Compiled skills are first-class: list the library and invoke any skill by id with typed inputs.
// List the library
const skills = await twin.skills.list();
// → [ book-slot@v3, export-report@v2, … ]
// Run one directly, no planner
await twin.skills.run("export-report", {
inputs: { range: "30d", format: "csv" },
});- Compiledone
- Organizerunning
- Reusequeued
- Iteratequeued
What agent & skill library is
The facts — how it works, what it costs, and the signal you get back on every call.
| Property | Twin Browser |
|---|---|
| Unit | Versioned skill |
| Invoke | By id, typed inputs |
| Surfaces | API, MCP, dashboard |
| Sharing | Reusable across agents |
| History | Runs + session video |
| Pinning | Pin to a version |
Agent & skill library — common questions
Can I run a skill without giving a goal?
How do versions work?
Can different agents share skills?
The rest of the platform
Deterministic replay
A successful run compiles into a skill — an ordered, parameterized program of browser actions that replays the same way every time.
Cross-tenant skill corpus
Common flows — log in, search, paginate, fill a form — get compiled once and shared through a sanitized corpus, so your agents start ahead.
Live view & session video
Stream the browser session in real time, then keep a durable video of every run for debugging, audit, and proof of what happened.
Put your agent to work — you stay in control.
Start free. Hand your agent a goal on any site you authorize, set the guardrails, and let it do the work — repeated runs compile into skills that replay at near-zero cost.