API
Splicora, for scripts and agents.
The same operations the app performs, through the same services and at the same prices: start a project from a link, find the clips in it, generate an image, a video, a voice or a track, publish it, schedule it. Over HTTP with a key you issue yourself, or as tools in an MCP client.
10 operations · 120 requests a minute per key · the same credits the app spends
- Operations
- 10
- Rate limit
- 120/min
- Keys per workspace
- 25
- Webhook events
- 6
The surface
Grouped by the job, not by the noun.
Every endpoint below is read from the router that serves it, so this list cannot name one that does not exist or miss one that does. The reference has the arguments and the answers.
Projects
A project is one recording and everything made from it. Start one from a link or an upload and read its state back.
-
POST
/api/v1/projectsCreate a project from a URL -
GET
/api/v1/projects/{id}Read a project
Clips
Ask for the short clips in a project. The answer is a job; the clips arrive with their scores and timings.
-
POST
/api/v1/projects/{id}/clipsFind the short clips in a project
Generation
An image, a video, a voice-over or a track from a prompt. Credits are reserved up front and refunded in full if it fails.
-
POST
/api/v1/generateGenerate an image, a video, a voice-over or a track -
GET
/api/v1/jobs/{id}Read a job
Publishing
Send a finished item to a connected account now, or put it on the calendar. The library lists everything the workspace has made.
-
POST
/api/v1/publishPublish to a connected account -
POST
/api/v1/scheduleSchedule a post for later -
GET
/api/v1/libraryList everything made in this workspace
Webhooks
Subscribe an endpoint of your own and hear when a render finishes or a post goes out, instead of polling for it.
-
POST
/api/v1/webhooksSubscribe to events -
DELETE
/api/v1/webhooks/{id}Stop sending to an endpoint
Keys
One key, one workspace.
Make a key in Settings, API keys, once you are signed in.
It starts splk_, it is shown once and stored as a hash, so copy it
when you make it. A key acts on the workspace it was made in and spends that workspace's
credits, which is what makes a key per client safe to hand out. Revoking is immediate, and a
workspace can hold 25 live keys.
Send it as a bearer token. A missing, wrong or revoked key all
answer 401 with the same body, on purpose.
curl -X POST https://splicora.ai/api/v1/generate \
-H "Authorization: Bearer splk_your_key_here" \
-H "Content-Type: application/json" \
-d '{"kind":"image","prompt":"a cold brew on a windowsill"}'
# { "jobId": "...", "credits": 4 }
In JavaScript
Ask, then wait for the job.
Anything that takes more than a moment answers with a job rather
than holding the request open. By the time you have the job id the credits are reserved and
the work is under way. Poll until done, then read succeeded. A job
that failed has already been refunded in full.
const key = process.env.SPLICORA_API_KEY;
async function call(method, path, body) {
const res = await fetch("https://splicora.ai/api/v1" + path, {
method,
headers: { Authorization: "Bearer " + key, "Content-Type": "application/json" },
body: body ? JSON.stringify(body) : null,
});
const data = await res.json();
if (!res.ok) throw new Error(data.error);
return data;
}
const { jobId } = await call("POST", "/generate", {
kind: "image",
prompt: "a cold brew on a windowsill",
});
let job = await call("GET", "/jobs/" + jobId);
while (!job.done) {
await new Promise((r) => setTimeout(r, 2000));
job = await call("GET", "/jobs/" + jobId);
}
Webhooks
Or do not poll at all.
Subscribe an endpoint of your own and name the events you want. Nothing is sent for an event you did not name. Each delivery carries a signature header, and the reference shows the check to make against it.
job.completedjob.failedclip.readypublish.succeededpublish.failedautomation.held
MCP
The same operations are available to an agent as tools, over
streamable HTTP at https://splicora.ai/mcp, behind the same key and the same
limit. One list, two protocols: a tool cannot exist in one and not the other.
The document
The whole surface as OpenAPI, generated from the schemas the routes validate with, at /api/v1/openapi.json. Open it in whichever client you already use.
What it costs
The same credits, at the same prices.
There is no separate API plan and no per-call fee. A call spends the credits from the workspace's monthly allowance, at the price the app charges for the same action, and the answer to a generation tells you what was reserved. A failure refunds it in full. The pricing page has the allowance on each plan and what every action costs.
The limit is 120 requests a minute per key, which is about throughput rather
than about your plan: credits are what meters the work. A 429 carries
Retry-After in seconds.
Start with a key.
Free plan, no card. The reference has every argument.

