On this page
Connect an agent
StrikeFree speaks MCP (Model Context Protocol), so Claude, Cursor and any MCP client can create jobs, wait for them and read the cue sheet with their own tools.
Remote server (streamable HTTP)
Point the client at the API with your key as a Bearer token. No install.
claude mcp add --transport http strikefree https://api.strikefree.video/mcp \
--header "Authorization: Bearer sf_live_..."
{
"mcpServers": {
"strikefree": {
"url": "https://api.strikefree.video/mcp",
"headers": { "Authorization": "Bearer sf_live_..." }
}
}
}
Local binary (stdio)
strikefree-mcp runs on your machine and can upload local files. It talks to the API with the key in its environment.
{
"mcpServers": {
"strikefree": {
"command": "strikefree-mcp",
"env": { "STRIKEFREE_API_KEY": "sf_live_...", "STRIKEFREE_API_URL": "https://api.strikefree.video" }
}
}
}
Tools
| Tool | What it does |
|---|---|
| submit_job | Create a job from a URL (or a local path with the stdio binary). Options music, keep, min_similarity, max_songs, duck_db, name, webhook_url. Returns the job. |
| get_job | Read a job; optional wait_s long-polls until it finishes. Done jobs carry the cue sheet and output links. |
| list_jobs | Recent jobs, newest first. |
| cancel_job | Cancel a queued or running job. |
| get_account | Plan and credits (balance, held, available), so the agent knows its budget. |
| download_output | Save an output to a local path (stdio binary only). |
For any other agent
GET /llms.txt is a one-page guide written for language models, and GET /openapi.json is the full contract. Both are public.
curl https://api.strikefree.video/llms.txt
curl https://api.strikefree.video/openapi.json