Sirv REST API

Base URL: https://api.sirv.com

Official Sources First

Check current Sirv docs before changing endpoint fields, path escaping, metadata field names, or pagination behavior:

• https://apidocs.sirv.com/
• https://sirv.com/help/articles/search-files-with-the-api/
• https://sirv.com/help/articles/add-meta-with-the-api/
• https://sirv.com/help/articles/sirv-api/

Authentication

All requests require a Bearer token from /v2/token:

curl -X POST https://api.sirv.com/v2/token \
  -H "Content-Type: application/json" \
  -d '{"clientId": "YOUR_CLIENT_ID", "clientSecret": "YOUR_CLIENT_SECRET"}'

Response:

{"token": "eyJhbG...", "expiresIn": 1200, "scope": ["account:read", ...]}

Use token in subsequent requests:

curl https://api.sirv.com/v2/account \
  -H "Authorization: Bearer eyJhbG..."

Tokens expire in 20 minutes. Request a new one before expiry.

Debugging Workflow

1. Identify the exact endpoint and method in play.
2. Reproduce or inspect the request body, query string, and response body.
3. Compare the request shape against the official docs before patching code.
4. Separate auth/token failures from Sirv payload/query failures.
5. Preserve upstream status and useful Sirv error detail when surfacing failures.
6. Add a focused regression test or fixture for the exact endpoint/query that failed.

Operational Rules

• URL-encode file and folder paths in query strings, especially / as %2F.
• Refresh tokens in long-running scripts; do not assume a token survives a bulk migration.
• Check /v2/account/limits before bulk search/upload/delete jobs.
• Use /v2/files/fetch to import remote originals directly into Sirv when source URLs are stable.
• Preserve catalog context with metadata after upload: title, description, tags, product fields, and approval state.
• Search returns up to 100 results per page. Use from for normal pagination and scrolling search for more than 1000 results.
• Scrolling search is a point-in-time snapshot and is cached for about 20 minutes; download results promptly.
• Escape search special characters in paths: { } / \ ! space.

Quick Reference

File Operations

Metadata Operations

Async Jobs (return job ID, poll for progress)

When to Read Reference Files

• File operations (upload, download, copy, delete, directory listing): See files.md
• Metadata & search (meta fields, search query syntax, product data): See metadata.md
• Async jobs (video conversion, ZIP, batch ops): See jobs.md
• Account & stats (usage, billing, events, settings): See account.md

Common Patterns

Upload an image

const token = await getToken();
await fetch('https://api.sirv.com/v2/files/upload?filename=/images/photo.jpg', {
  method: 'POST',
  headers: {
    'Authorization': `Bearer ${token}`,
    'Content-Type': 'image/jpeg'
  },
  body: imageBuffer
});

Search for recent images

await fetch('https://api.sirv.com/v2/files/search', {
  method: 'POST',
  headers: {
    'Authorization': `Bearer ${token}`,
    'Content-Type': 'application/json'
  },
  body: JSON.stringify({
    query: 'extension:.jpg AND mtime:[now-7d TO now]',
    size: 50
  })
});

Search a folder with escaped path

await fetch('https://api.sirv.com/v2/files/search', {
  method: 'POST',
  headers: {
    'Authorization': `Bearer ${token}`,
    'Content-Type': 'application/json'
  },
  body: JSON.stringify({
    query: 'dirname.paths:\\/products AND extension:.jpg',
    size: 100
  })
});

Create ZIP archive (async)

// Start job
const { id } = await fetch('https://api.sirv.com/v2/files/zip', {
  method: 'POST',
  headers: { 'Authorization': `Bearer ${token}`, 'Content-Type': 'application/json' },
  body: JSON.stringify({
    filenames: ['/images/photo1.jpg', '/images/photo2.jpg'],
    zipFilename: '/downloads/photos.zip'
  })
}).then(r => r.json());

// Poll until complete
let progress = 0;
while (progress < 100) {
  const status = await fetch(`https://api.sirv.com/v2/files/zip?id=${id}`, {
    headers: { 'Authorization': `Bearer ${token}` }
  }).then(r => r.json());
  progress = status.progress;
  await new Promise(r => setTimeout(r, 1000));
}

Red Flags

• Changing Sirv request syntax from memory instead of current docs.
• Flattening Sirv 400/401/403/429/5xx responses into a generic local error.
• Treating search field names, path escaping, or page-size limits as generic Lucene behavior.
• Fixing token acquisition and request body shape in the same step without proving which boundary failed.
• Hiding the exact Sirv path, folder, tag, or payload involved in a failure.

Verification

• Reproduce the failing request with the exact folder, tag, file path, or payload.
• Confirm file paths are URL-encoded in query strings and escaped in search syntax where required.
• Confirm page size, pagination, or scrolling matches Sirv's documented behavior.
• Verify imported/uploaded files with stat, meta, readdir, or search.
• Re-run the relevant integration/route test when this skill is used in a codebase.