One key. One endpoint.
Connect text and data utilities to n8n, Make, Zapier or your own application. The hosted beta supports 44 operations with verified accounts, API keys and usage tracking.
Make your first request
Download the n8n text and file workflow. Set your reachable API URL, private Header Auth credential and writable output directory before running it.
Create an account, verify your email, then create an API key. Store it in your integration's secret credentials.
POST https://pixmello-web-api.onrender.com/api/v1/execute
Authorization: Bearer YOUR_KEY
Content-Type: application/json
{"tool":"sha256-hash","input":"hello"}The hosted API base URL is https://pixmello-web-api.onrender.com. Set PIXMELLO_BASE_URL to this origin in the examples below. Do not put keys in query strings or browser code.
In n8n
- Add an HTTP Request node; select POST.
- Set your Pixmello host followed by
/api/v1/execute. - Choose Generic Credential Type → Header Auth. Name:
Authorization. Value:Bearer YOUR_KEY. - Send a JSON body with
tool,inputand optionaloptions. - The output is in
result.text. Useresult.mimeandresult.filenamewhen saving it.
Make and Zapier can use their HTTP request modules with the same headers and JSON body.
44 supported operations
OpenAPI document · Machine-readable capabilities · Local SDK documentation
Every request uses POST /api/v1/execute. No separate upload, job, webhook or pipeline endpoints exist in this release.
Search API operations
SHA-256 hash generator
Calculate a SHA-256 digest of UTF-8 text.
Limits: Hashes the exact UTF-8 input including spaces and newlines. Empty text has a valid digest. Up to 256 KiB. Raw SHA-256 is not a password-storage scheme.
{
"tool": "sha256-hash",
"input": "abc",
"options": {}
}cat > request.json <<'PIXMELLO_REQUEST'
{
"tool": "sha256-hash",
"input": "abc",
"options": {}
}
PIXMELLO_REQUEST
curl "$PIXMELLO_BASE_URL/api/v1/execute" \
-H "Authorization: Bearer $PIXMELLO_API_KEY" \
-H 'Content-Type: application/json' \
--data-binary @request.jsonAuthenticated API playground
Uses your real key and quota. Key, request and response stay in this page's memory; they are not saved to browser storage. Cancellation stops waiting, not an already accepted server operation.
Response formats
Every success contains requestId, tool and result. Text tools return result.text, MIME and filename. Binary tools return kind: file, encoding: base64, data, decoded bytes, MIME, filename and measured metadata. ZIP extraction returns a JSON text manifest; decode each entry's Base64.
{
"requestId": "UUID",
"tool": "image-converter",
"result": {
"kind": "file",
"encoding": "base64",
"data": "PADDED_BASE64_BYTES",
"bytes": 1234,
"mime": "image/png",
"filename": "converted.png",
"metadata": {
"metadataStripped": true
}
}
}This is a response-shape example, not a measured processing result.
Authentication, quotas and file requirements
Bearer keys are issued only to verified accounts. Use one per integration; never put keys in URLs or public client code. Key revocation and expiry apply on every request. 10 requests/minute, 100/day and 1,000/month per account, shared across keys. UTC windows. Shared platform budget: 300/day and 10,000/month. Keys expire after 90 days.
JSON requests only; no multipart or remote URLs. Maximum request body: 2,000,000 bytes. File inputs use padded standard Base64; combined decoded inputs at most 1 MiB. At most 10 PDF pages/4 million image pixels. Binary outputs at most 1 MiB. Each instance runs two file-engine workers. Up to eight attempts can wait for at most 1.5 seconds for admission; excess attempts return 503. Workers have a 12-second deadline and 192 MiB JavaScript heap; native allocations are additionally bounded by pixel/file limits, not a hard RSS sandbox.
Pre-admission capacity denials do not consume execution quota. Once admitted, engine attempts count even when processing fails. Invalid auth/request envelopes do not consume processing quota. No automatic paid upgrade or overage service is enabled. Valid processing results survive delayed usage-status logging; administrative maintenance labels stale receipts as unknown outcomes rather than inventing success or failure.
Files, inputs and outputs are transient; they are not saved to PostgreSQL or disk. Async jobs, callbacks, remote fetching and idempotent retries remain unavailable. Website tools keep their browser-first engines.
Error diagnostics and retries
Errors use {"error":{"code":"…","message":"…"}}. Keep the returned request ID for requests admitted to processing; never include input files or keys in support messages.
| HTTP | Codes | Action |
|---|---|---|
| 400 | INVALID_INPUT, INVALID_JSON, IDEMPOTENCY_UNAVAILABLE | Fix the request; idempotency headers are rejected. |
| 401/403 | INVALID_API_KEY, SIGN_IN_REQUIRED, EMAIL_VERIFICATION_REQUIRED, ACCOUNT_DISABLED, INVALID_ORIGIN | Verify identity/credentials and resource ownership. |
| 413/415 | PAYLOAD_TOO_LARGE, LIMIT_EXCEEDED, INVALID_CONTENT_TYPE | Use smaller valid JSON/file inputs. |
| 429 | RATE_LIMITED, AUTH_RATE_LIMITED, QUOTA_EXCEEDED, PLATFORM_QUOTA | Respect Retry-After when present and the UTC quota reset; do not retry in a tight loop. |
| 503 | CAPACITY_EXCEEDED, PROCESSING_TIMEOUT, PROCESSING_UNAVAILABLE, SERVICE_UNAVAILABLE, SETUP_REQUIRED | Back off. An admitted failed attempt still counts; retries are not deduplicated. |