List streams in your WAVE account (GET /v1/streams), with pagination and status filtering

Create a new stream (POST /v1/streams) with a Cloudflare Stream live input

Start a stream by its ID (POST /v1/streams/{id}/start)

Stop an active stream by its ID (POST /v1/streams/{id}/stop)

Get the current status document for a stream (GET /v1/streams/{id}/status): connection state, viewer count, and quality indicators

Get analytics for a single stream (GET /v1/streams/{id}/analytics): views, watch time, and geographic/device breakdown over an optional date range

Mark a moment in a stream as a highlight for later clipping (POST /v1/streams/{id}/highlights)

List multi-camera productions in your WAVE account (GET /v1/productions)

Create a new multi-camera production (POST /v1/productions)

Get account-wide viewer engagement analytics (GET /v1/analytics/engagement) over an optional date range

Get the current billing account (GET /v1/billing): plan, Stripe customer status, and subscription state

Get billed usage for a date range (GET /v1/billing/usage), defaulting to the current month

Switch the program or preview bus to a different camera index in a multi-camera production (POST /v1/productions/{id}/camera)

Create a clip from a recording (POST /v1/clips). source is a recording ID; in and, optionally, out/duration are time strings like "5s" or "2m"

Show or hide a graphics overlay in a production (POST /v1/productions/{id}/overlay)

Send a control command to a managed camera (POST /v1/cameras/{id}/control). command selects the variant; supply only the fields that variant needs: set_iris/set_focus/set_zoom/set_gain use value; set_white_balance uses temperature+tint; set_shutter uses angle; set_audio_level uses channel+level; recall_preset uses preset_id; save_preset uses name+slot; start_recording/stop_recording/start_prerecord/autofocus_trigger take no extra fields

Moderate a chat message in a live stream (POST /v1/moderate): block, flag, or allow it

Transcribe an audio clip and optionally run a fast-LLM step over the transcript (POST /v1/live/pipeline, multipart). This processes ONE provided audio chunk through WAVE's live pipeline — it does not attach a persistent caption feed to a live stream. Requires the account's live pipeline feature to be enabled; a 404 means it is not armed for this account

Drive a full headless conversation with the WAVE voice agent: bind the agent to a room, send a WAV of the caller's speech (16-bit LE 48 kHz PCM, mono or stereo), and receive the agent's spoken reply as raw 16-bit LE 48 kHz stereo PCM written to outPath. No browser, no WebRTC.

Run @wave-av/pen-extract's all extraction pipeline on a .pen board (plates/tokens/geometry/copy/components/receipts/frames/manifest). Returns the manifest.json (files, sha256s, owed).

Compose a design-contract.json from an already-extracted pen-extract dir, then validate it against the schema + acceptance-tests catalogue. Returns the validator line, key counts, and the written path.

Run @wave-av/loc-study's measure on an image (masked by geometry.json when given) or a rasterized plate SVG, optionally cropped to a region. Returns the measurement JSON (bokashi/keyblock/palette/composition).

Validate an existing design-contract.json against the schema + acceptance-tests catalogue, no compose.

Propose a WAVE flow (product + MCP tool + price shape) for a media-processing goal stated in plain language (captions/clips/dub/realtime/identity/x402/...). Reads only: checks a bundled, measured snapshot of the product/skill/tool manifests (knowledge/*.json) — never the live network. Never executes: calls no other tool, signs no payment, and makes no side-effecting or network request. Returns a proposal object for you to call yourself; always carries the literal executes: false and never a model field (no sourced Dispatch model catalog exists). See skills/wave-ask/SKILL.md for the full contract.