Glossary · Cost engine

Skill compilation

Turning a successful agent run into a reusable, replayable skill — a compact program of indexed actions you can call again without re-planning.

Start free All terms

What is skill compilation?

Skill compilation is the step where a successful, LLM-planned run is distilled into a durable skill: an ordered set of actions bound to token-efficient DOM indices plus the inputs it needs. The expensive reasoning happens once during this cold compile; every later invocation can replay the skill or match it through the semantic cache.

Why it matters

Compilation is the moment cost moves from per-run to one-time. It separates the act of figuring out a workflow (model-heavy) from running it (model-free), which is exactly the boundary RPA-replacement and high-volume agent teams need to make per-1k-run cost fall.

See it in context: read how Twin compiles and replays a run, follow the cost-cutting guide, or browse use cases and comparisons.

Related terms

Deterministic replay

Re-running a compiled skill as a fixed sequence of indexed actions, with zero LLM calls, so the same goal produces the same path every time.

Semantic dispatch cache

A vector cache that matches a new, re-phrased request to a skill you already compiled — so a repeated task skips the LLM.

Cross-tenant skill corpus

A shared, sanitized library of compiled skills that lets a skill compiled once be safely reused across tenants, compounding the cache savings.

Browser execution layer

The infrastructure that turns an agent’s goal into deterministic, replayable browser action — the layer between your LLM and a real browser.

Back to the full glossary

Run your first skill

Give an LLM agent a real browser, compile the workflow once, and watch the marginal cost fall as the cache takes over.

Start free See pricing