Building with an AI assistant?

Generate a ready-made prompt tailored to Cursor, Claude Code, ChatGPT, or any LLM.

PRIMARY AUTOMATION · START HEREtask runsone goal, driven to doneworkflowsmany tasks, one programmachinesmanaged Linux or Windowsuse finer control when neededPRIMITIVES · YOU OWN EXECUTIONpredictstateless stepsessionsstateful loopgroundfind coordinatesparsecode to actions

Start with task runs, workflows, and machines. Drop to prediction primitives only when your application needs to own the control loop.

Start with a task run: give Coasty a goal and a machine, then let the agent drive to completion. Use workflows when an automation needs many tasks, branches, loops, approvals, or shared outputs. Machines provide the managed computer those tasks operate.

The prediction endpoints are lower-level primitives for teams that need to own the control loop themselves. Use sessions for a stateful screenshot loop, predict for a stateless step, grounding for coordinates, and parse for structured actions. Everything is normal HTTPS to https://coasty.ai/v1, so you can choose the highest-level surface that fits the job and drop down only when you need finer control.

API keyscope ✓reserve $run model200 + usageX-Credits-*failsrefund + 5xxX-Credits-Refunded401 / 403 / 402 short-circuit before any charge

Every billed call is charged before the model runs and automatically refunded if it fails — the X-Credits-Refunded header confirms the refund.

Every API-key-authenticated request must include your secret key. The four health probes are public, and webhook ingress uses its documented Coasty-Signature HMAC credential instead. For API-key operations, the canonical form is the X-API-Key header, but Authorization: Bearer <key> works too: a blank X-API-Key falls through to the Bearer header. Pick one form and send the raw key. Do not paste the literal text Bearer  inside X-API-Key; that is the single most common first-day mistake and it returns 401 INVALID_API_KEY. Keys are created and revoked from the API keys page. Treat a key like a password: keep it server-side, store it in an environment variable, and never commit it or ship it in client-side code.

Prefer test keys while you wire up your integration. An sk-coasty-test-

key never bills and runs against mock VMs, yet exercises the exact same request and response shapes (its X-Credits-Charged and usage.cost_cents

are always 0), so you can build and run CI confidently before flipping to a live key.

Your first autonomous task needs an API key and a machine. Grab a test key from the API keys page (it never bills), then use an existing machine or provision one. Set the key in your shell:

Start the task with POST /v1/runs. Replace the example machine_id, describe the outcome in task, and send an Idempotency-Key so a retried create cannot start a duplicate run. The complete example starts the run and follows it to a terminal state:

The create response begins at queued. Coasty then drives the machine, records each step, and finishes as succeeded, failed, cancelled, or timed_out. Your application can poll, subscribe to the event stream, or receive signed webhooks; it does not need to execute each prediction itself.

Task runs are the default starting point for autonomous work. Workflows build on them, machines host them, and prediction primitives remain available when your application needs direct control over the loop.

POST /v1/runstask + machineagent loopseescreenshotthinkpredictactapplyverifydonepass / failSSE eventslive stream, resumablewebhookHMAC-signed callbackshuman takeoverpause → resume

POST a task + machine and Coasty runs the whole loop for you — streaming events over SSE, calling your webhook on lifecycle changes, and pausing for a human when asked.

A run hands the agent a task and a machine, then drives it to completion on our side. The agent loops autonomously, verifies its own work (pass or fail), can pause for a human when it hits a wall, bills $0.05 per completed step from your dollar API wallet ($0.08/step on the legacy v1 engine), and streams every event live. You start one call and watch, instead of running the predict loop yourself.

Create a run with POST /v1/runs. The two required fields are machine_id and task. The response is an agent.run object with status of queued, plus a one-time webhook_secret you store to verify webhooks. Send an Idempotency-Key header to make a retried create safe.

A run moves through queued to running, can bounce between running and awaiting_human, and ends in one of succeeded, failed,

cancelled, or timed_out. Terminal states are immutable, so it is always safe to stop polling once you reach one. Runs need the

runs:read and runs:write scopes, granted to new keys by default.

GET /v1/runs/{id}/events returns a Server-Sent Events stream so you can follow a run as it happens, instead of polling. Each event has a type and a numeric id (the sequence number). If your connection drops, reconnect and replay everything you missed by sending the last sequence you saw as a Last-Event-ID header, or as the ?after= query parameter. The stream closes after the done event.

Some steps need a person: a captcha, a one-time code, a judgment call. When the agent reaches one and on_awaiting_human is pause, the run moves to awaiting_human and emits an awaiting_human event with a reason. A human completes the blocking step (in the same machine session), then you hand control back with POST /v1/runs/{id}/resume and an optional note. Resume is only valid while the status is awaiting_human.

Detect the pause from either the run object (status == awaiting_human

with awaiting_human_reason set), the SSE

awaiting_human event, or the run.awaiting_human

webhook. After resume, the run returns to running and emits a

resumed event. Set on_awaiting_human to

fail or cancel at create time if you would rather the run stop than wait for a human.

Pass a webhook_url (https only) when you create a run and we POST a signed callback at each lifecycle transition. The response to your create call includes a webhook_secret exactly once: store it, because every callback is signed with it. Each request carries a Coasty-Signature header of the form t=<unix_ts>,v1=<hex>.

To verify, build the signed payload as "<t>." + raw_request_body, compute HMAC-SHA256 over it keyed by the webhook_secret, and compare against v1 with a constant-time check. Always hash the raw body bytes, before any JSON re-serialisation.

By default every LLM call in the computer-use harness runs on Coasty's managed models. BYOK (bring your own key) flips that: opt in and the entire harness (the worker, grounding, the code agent, and compaction; every LLM call) runs on your own Anthropic or OpenAI account instead. Opt-in is always explicit, per request or per stored key. provider: "managed" (or omitting llm entirely) keeps the platform default, unchanged.

There are two ways to hand over a key. Store it once with PUT /v1/llm/keys/{provider} (encrypted with AES-256-GCM at rest; only a sha256-prefix fingerprint is ever echoed back), or send it per request in headers. A header key takes precedence over the stored key. Sending a key without a provider returns 422.

The headers work on POST /v1/predict, POST /v1/runs, POST /v1/workflows/{id}/runs, POST /v1/sessions, and POST /v1/schedules. Stored keys are managed through three endpoints, gated by the llm_keys scope (granted to new live keys by default). These endpoints require a live key; sandbox keys cannot read, overwrite, or delete the production credential store:

On the request body, the same endpoints accept an llm object that selects the provider and, optionally, a model per harness role. It deliberately has no api_key field (a 422 if you attempt one): keys ride headers or the encrypted store only, so they can never be echoed in run objects, webhooks, or idempotency replays.

Per-role overrides exist for tuning cost against quality: run compaction on a cheaper model while the worker stays on the default, for example. For runs, workflows, and schedules the key is snapshotted encrypted into the run, so crash-recovery on another replica keeps using your key; it is scrubbed the moment the run reaches a terminal state. GET /v1/runs/{id} echoes a non-secret llm block: {provider, model, key_fingerprint, key_source, key_scrubbed}. Schedules store only the non-secret preference; at fire time the current stored key is used. Delete the stored key and future firings fail loudly with LLM_KEY_NOT_CONFIGURED; they never silently run on platform keys.

Grounding (pixel-coordinate resolution) quality is tuned on the platform model. When running grounding on your own model, expect the best results with the defaults, and use

grounding_model to experiment before committing a cheaper or different model to that role.

No silent fallback, ever: once you ask for BYOK, no code path can use Coasty's platform LLM keys. Errors from your provider surface as your key's issue with stable codes:

LLM_KEY_NOT_CONFIGURED, LLM_KEY_INVALID,

LLM_PROVIDER_AUTH_FAILED, LLM_PROVIDER_RATE_LIMITED,

LLM_PROVIDER_QUOTA_EXCEEDED, and LLM_PROVIDER_ERROR.

Billing: Coasty's per-call and per-step platform charges are unchanged with BYOK. The model tokens are billed by your provider account directly.

workflowversioned JSONtask= a runassertpureif / looppuretask= a runbudget_cents · max_iterations · deadline guard the whole run

A workflow is a small JSON DSL. Control steps (if / loop / parallel / human_approval) are pure and safe; each task step executes as a real run with its own billing and events.

A workflow composes many runs into one versioned program, with branching, loops, and guards expressed as a JSON DSL. Each task step is itself an agent run, so a workflow is the way to chain tasks, gate them on conditions, and pass results between them. Workflows are versioned: re-creating the same slug bumps the version, and a PUT does too.

Create one with POST /v1/workflows. The slug must match [a-z0-9_-]. The response is a Workflow carrying an id, a version, and the current dsl_version (2026-06-01).

Workflows need the workflows:read and

workflows:write scopes, granted to new keys by default. See the

Workflow DSL

for the full step and condition catalogue.

The DSL (dsl_version 2026-06-01) is a JSON object with a steps array and an optional output. Each step has an id and a type. A task step runs the agent and binds its result ({ status, passed, result, run_id, steps, error }) under both its save_as name and its step id, so later steps can read it.

Conditions are structured rather than expression strings, which keeps them injection-safe. Each left, right, or value is either a literal or a {{path}} reference. Paths are dotted lookups into inputs.*, vars.*, and any step id or save_as name.

Three hard guards stop a workflow run when breached:

budget_cents (spend cap in USD cents; 0 means unlimited),

max_iterations (loop cap), and

deadline_seconds (wall-clock). A breach ends the run as

failed or timed_out.

A definition is validated before it is accepted. The limits below are enforced at create and ad-hoc time, so an invalid definition is rejected with 422 VALIDATION_ERROR rather than failing mid-run.

Workflows are version-pinned. When a run starts, the workflow's current

definition is snapshotted into that run, so editing or replacing the workflow (which bumps its version) never changes runs already in flight. Each run records the workflow_version it executed.

Start a saved workflow with POST /v1/workflows/{id}/runs, or run a definition inline (without saving) with POST /v1/workflows/runs by adding a definition (and optional inputs_schema) to the same body. Both return a workflow.run. The body accepts inputs, a default machine_id for task steps, and the budget_cents, max_iterations, and deadline_seconds guards. An Idempotency-Key header is honoured here too.

A machine is a Coasty-managed cloud VM that an agent can see and control. You provision one, poll until it is running, then either hand it to a task run (the agent drives it to done) or drive it yourself with the action endpoints. Machines are optional: /predict, /sessions, and /ground run against your screen. You only need a Coasty machine when you want the agent to execute on a VM Coasty hosts.

Provision with POST /v1/machines. Only display_name is required; everything else has a sensible default. The body rejects unknown fields, so a typo returns 422 VALIDATION_ERROR rather than being silently ignored.

Provisioning is asynchronous. The response returns immediately with the machine in creating status and a connection object whose secrets are redacted. The VM is not drivable yet — poll GET /v1/machines/{id} until status is running before you send actions or start a run.

The machine object — returned by provision, list, and get:

List your machines with GET /v1/machines (newest first, ?limit= 1–200, default 50) — it returns { data, has_more, request_id }. Fetch one with GET /v1/machines/{id}. Both read straight from the registry, so they keep working even when provisioning is busy.

Provisioning a live machine needs the machines:write scope, a wallet balance of at least 20 credits (Unlimited-tier keys skip this), and room under your plan's concurrent-machine cap. Building? An sk-coasty-test- key returns a fully-shaped mock VM (id mch_test_…, is_test: true) with zero billing — up to 5 at a time.

creating0 cr/hrrunning5–20 cr/hrstopped1–2 cr/hrterminated0 cr/hrstopstartdelete / TTLttl_minutes elapsesinsufficient funds → machine is stopped & flagged, never deleted — your data is safe

Machines bill per minute while they exist. Out of funds → stopped (never destroyed). A TTL auto-destroys the VM at created_at + ttl_minutes.

A machine moves through a small set of statuses. status is the field you poll: drive the machine only while it is running. The runtime rate that applies in each status is shown below; exact per-hour USD numbers are in the Pricing section.

Start, stop, restart are POST /v1/machines/{id}/start (and /stop, /restart). They are asynchronous: the call returns a transitional status (starting / stopping) and you poll until it settles. They are state-checked — starting a machine that is already running, or stopping one that is not running, returns 409 INVALID_STATE with current_state and allowed_from in the body, so you can react without guessing. start is allowed from stopped or error; stop only from running.

Terminate with DELETE /v1/machines/{id}. This is permanent — the VM and its disk are destroyed and a later GET returns 404 MACHINE_NOT_FOUND. Delete is idempotent: deleting an already-gone machine still succeeds, so retries are safe.

Auto-destroy (TTL). A machine left running bills until you destroy it, so set ttl_minutes as a safety net. A background sweep (every ~60s) terminates a machine once its auto_destroy_at passes. Adjust it any time with PATCH /v1/machines/{id}: ttl_minutes is measured from now (so it doubles as a lease extension), accepts 5–10080 (5 min to 7 days), and 0 clears auto-destroy entirely. Anything else is 400 INVALID_TTL. Provisioning without a TTL works but adds a Warning response header nudging you to set one.

Runtime billing. Machines bill the developer API wallet (separate from any subscription credits) by the minute, rounded down to whole credits in your favour. Running Linux is $0.05/hr, running Windows $0.09/hr, and a stopped or suspended machine the keep-alive rate of $0.01/hr; creating, error, and terminated are free. Transitional states (starting/stopping/restarting) bill at the running rate. The per-call control endpoints (actions, terminal, files, browser, screenshot, connection) are never billed — you pay for runtime only.

If the API wallet empties while a machine is running, the next sweep stops it (status becomes suspended) rather than destroying it — your disk and snapshots are preserved. Top up the wallet and POST /start to resume. You can watch live accrual for every metered machine at GET /v1/billing/active.

1st requestIdempotency-KeyreserveSET NXexecutecharge oncestore result24hretry (same key)timeout / 502replays the stored result — X-Credits-Charged: 0

On the exact 18 reserve-and-replay operations, send Idempotency-Key with the original request. A retry with that same key replays the stored result and never bills twice; unsupported mutations cannot gain idempotency retroactively.

Once a machine is running you can reach it three ways: connect directly over SSH/VNC, drive it through Coasty's control endpoints, or hand it to an agent run. Normal machine responses redact secrets and only expose has_ssh_key / has_vnc_password booleans plus ports.

Connection secrets. GET /v1/machines/{id}/connection returns the full ssh_private_key_pem, vnc_password, public IP, and ports. It is gated by the opt-in connection:read scope (not granted by default — request it when you mint the key), and the response is sent Cache-Control: no-store. Treat that payload like a password: never log it. SSH usernames are ubuntu on Linux and Administrator on Windows; VNC details exist only on desktop_enabled machines.

Screenshots & snapshots. GET /v1/machines/{id}/screenshot returns the current screen of a desktop machine (a still-booting VM returns 502 SCREENSHOT_FAILED — poll for running first). POST /v1/machines/{id}/snapshot captures a restorable image (Linux), needs the snapshots:write scope, and is the one machine op with a flat fee ($0.01 per snapshot). A conclusive pre-creation failure is refunded, confirmed by X-Credits-Refunded. A timeout, an upstream 5xx, or malformed post-dispatch result may already have created the image and instead returns terminal 503 SNAPSHOT_OUTCOME_UNKNOWN without a blind refund or re-execution. Boot a future machine from a confirmed snapshot with restore_from_snapshot: true.

The control surface. Drive the VM directly with these endpoints. Each enforces a specific scope, and the high-risk ones — browser_execute (arbitrary JS) and anything under connection:read — are opt-in. /actions/batch runs up to 50 steps and stops on the first error by default (stop_on_error: false to continue); /terminal truncates output to 5000 chars.

Idempotency & safety. The machine operations in the exact reserve-and-replay set — provision, snapshot, actions, action batches, browser, terminal, and file operations — accept an Idempotency-Key (≤128 chars). Lifecycle start/stop/restart, TTL updates, and termination do not. For supported operations, a duplicate key replays the original result without re-executing; a duplicate that arrives while the first is still running waits up to ~25s, then returns 409 IDEMPOTENCY_IN_FLIGHT. One nuance worth knowing: a command that never reached the VM (a dispatch failure) is not cached, so retrying the same key runs it fresh; a command the VM actually ran (even if it errored) is cached. Every machine lookup is ownership-scoped, so a wrong or someone-else's id returns 404 — never a leak. Full code list in Errors.

screenshotbase64 PNG / JPEGPOST /v1/predictCoasty CUAvision modelactions[]your appapply the actionsrepeat until status = "done"click · type_text · scroll · key_combo · drag · done

The low-level prediction loop: send a screenshot, apply the returned actions, then capture the next screen and repeat until status is done.

POST /v1/predict is the low-level, stateless prediction primitive. Each call is independent: you provide the full context, execute the returned actions, capture the next screenshot, and decide when to call again. Use it when your application must own that loop. For an autonomous goal, start a task run. When a manually driven loop needs server-side trajectory memory, reach for sessions instead.

The response is the standard prediction shape, covered in Response format.

A session keeps the trajectory — the running history of screenshots and actions — on our side, so each step only needs the latest screenshot and instruction. This produces better multi-step behaviour on long tasks and keeps your request bodies small. Create a session once, step through the task, then delete it to release your concurrency quota.

Always delete a session in a finally block. Sessions count against your tier's concurrent-session limit. The 2-hour (7200-second) idle TTL is reset on each predict and reset; deleting the session releases the slot immediately.

Grounding answers a narrower question than predict: “where is this element?” Give it a screenshot and a description and it returns the exact x, y coordinate to target. It is faster and cheaper than a full prediction ($0.03 instead of $0.05), which makes it ideal when you already know what to do and only need a pixel to click.

The response is { x, y, usage, request_id }. Coordinates are in the same pixel space as the screenshot you sent.

Parse converts a block of pyautogui code into the same structured action objects the model returns. It is deterministic, runs no model, and is free. Use it to migrate existing automation scripts onto Coasty's executor, or to normalise hand-written steps into the canonical action schema.

Every action the model can return uses an action_type from the table below, paired with a params object. Your executor switches on the type and applies the parameters. The terminal types — done and fail — set the response status and signal you to stop looping.

Predict and session-predict return the same shape. actions is the ordered list to execute; status tells you whether to keep going (continue), stop successfully (done), or stop because the task is impossible (fail). usage reports tokens and the dollar cost of the call (cost_cents).

Billed success responses also carry two headers you can read without parsing the body: X-Credits-Charged (what this call cost) and X-Credits-Remaining (your wallet balance after it). In the body, the same numbers appear as usage.credits_charged and usage.cost_cents. On an sk-coasty-test- key both are always 0. Every response (success or error) additionally carries an X-Coasty-Request-Id header that mirrors request_id; quote it when contacting support.

On a failed billed request, X-Credits-Refunded is the authoritative confirmation that credits were returned, and X-Credits-Charged is then0. If settlement cannot be confirmed, the API returns503 BILLING_UNAVAILABLE without the refund header. Follow the error'sretry_with_same_idempotency_key field rather than retrying by status alone.

Errors return a non-2xx status and a JSON envelope under an error key. The code is stable and safe to branch on; message is human-readable and may change. Every error also carries an error.request_id (mirrored in the X-Coasty-Request-Id response header), plus error.suggestion and error.docs_url for self-service. A Link: <url>; rel="help" header mirrors docs_url. Always log the request id: it is the fastest way for us to trace a failed call.

Some codes attach machine-readable context to the body. A 402 (INSUFFICIENT_CREDITS) reports required and balance; a 403 reports required_scope and current_scopes; a 422 VALIDATION_ERROR lists the offending field path under error.details; and a 409 state conflict carries current_state with allowed_from or required_state.

Treat 429, 503 (UPSTREAM_UNAVAILABLE), and 504 (UPSTREAM_TIMEOUT) as retryable: honor Retry-After

on a 429, and use an Idempotency-Key with exponential backoff on the upstream codes only when the response's retryable andretry_with_same_idempotency_key fields permit it. A500 model failure (PREDICTION_FAILED or GROUNDING_FAILED) is refunded only when X-Credits-Refunded is present; after a confirmed refund, use a new idempotency key for a new execution.

Troubleshooting

Five mistakes account for almost every first-week support ticket. Each maps to one status and one fix:

Requests are billed in US dollars from your prepaid API wallet. The charge is taken before the model runs. On a conclusive server-side failure, the API submits a refund; treat it as complete only when X-Credits-Refunded is present. Ambiguous provider mutations may retain the debit until reconciliation. Internally each request unit is $0.01 (the granularity behind every price below), but everything you pay and see is dollars. Every price on this page is exact; test keys (sk-coasty-test-) always bill $0.00.

Surcharges

Four fixed surcharges can apply on top of a base price, all on the vision endpoints (predict, session steps, ground). Each is an exact USD amount:

Machines

Machines bill for runtime only, metered per minute and rounded down: $0.05/hr for a running Linux machine, $0.09/hr for a running Windows machine, and $0.01/hr while stopped or suspended. The starting, stopping, and restarting transitions bill at the running rate; the creating, error, and terminated states bill nothing, and TTL auto-destroy is free. Snapshots are a one-time $0.01 each, and every per-call operation (actions, batch, browser, terminal, files, screenshot, connection) is free. Provisioning requires a $0.20 wallet minimum, which is a gate, not a charge. If the wallet empties mid-flight the machine is automatically stopped, never destroyed, and resumes after you top up. The live rate card is always at GET /v1/machines/pricing.

Schedules

Schedules have no per-fire fee: webhook fires are free (limited to 60/min), and create, run-now, and webhook fires only require the same $0.20 wallet minimum as a gate. The execution itself is billed differently from everything else on this page: scheduled agent runtime is charged to your subscription credit balance at 10 credits per minute ($0.10 of subscription value per minute, at 1 credit = $0.01), not to this USD API wallet. Keep both balances funded if you rely on schedules.

Curated open-source repositories for building on the Coasty API. Clone a repo, mint a key, and ship.