worca — the human layer, as a command
One CLI your agent calls to hand a task to a vetted human. Results print to stdout with proof attached, in minutes.
worca exposes Worca's owned, trained workforce in the Philippines behind a single command.
Your agent runs worca, a dedicated pod of vetted humans handles the work 24/7, and the
result returns on stdout with evidence.
What can a human do for your agent?
Four task types are live today. This is the first command your agent should run — it lists what can be delegated, straight from the source of truth:
worca tasks types
| Task type | What a human does | Proof returned |
|---|---|---|
social_post | Publishes your agent's content on a social platform, following exact steps | URL of the live post |
video_edit | Edits client-provided footage per a written brief — e.g. cuts a TikTok short | Link to the edited video |
phone_call | Makes a phone call following an exact script and playbook | Structured outcome report |
custom | Anything you can describe — Worca quotes a price, work starts after you approve | Outcome report + artifact link |
Each type has a strict input schema. worca tasks types <name> prints the required fields
(GET /v1/task-types over the wire), and input is validated pre-flight — bad requests fail
before any human is engaged. More: discover task types · full schemas in the
CLI and REST API references.
Who does the work
Work routes to a pod — a named group of trained humans, three shifts in Manila, 24/7. Production accounts get a dedicated pod that learns your workflow; 15-minute median SLA to first answer. Every result ships with proof you can verify.
Next: install · authenticate · quickstart · full command reference.