Skip to content

REST API

Call Super Artifacts from your own code over HTTPS with an API key — base URL, errors, pagination, retries, rate limits, key scopes and MCP parity.

reference~3 minVerified v0.175.0

Every MCP tool has an HTTPS endpoint. See API endpoints for operations, or use an SDK.

Base URL and key

Send JSON to https://superartifacts.app/v1. Create a key in Settings → API keys; send Authorization: Bearer <key>. The public OpenAPI document is https://superartifacts.app/v1/openapi.json.

Keys allow reading, or reading and writing. A read key writing gets 403 scope_required; older keys allow both. Keep keys server-side: requests carrying Origin are refused with 403.

Errors

Failures use {error: {code, message}}. Error codes lists stable codes.

Statuscode
400invalid_arguments, invalid_*
401unauthorized
402plan_limit
403forbidden, scope_required
404not_found
409conflict
413payload_too_large: over 1 MiB; use /v1/uploads
429rate_limited, workspace_busy
500internal

Lists

Arrays name their contents: artifacts, responses, entries. Data pagination returns cursor.next; send ?cursor=. Collection entries use ?page=.

Retries

A POST with Idempotency-Key replays its first answer for 24 hours. Artifact/project deletion requires ?confirm= matching its id.

Polling and fair use

Cache configuration in my workflow

Edge budgets per key: 600 reads/minute, 120 writes/minute, 20 email calls/hour. Workspace collection reads: 60/minute per collection per caller account, shared across keys. Artifact collection reads: 60/minute per collection per token holder within the artifact; collection writes: 30/minute per holder. Response writes: 30/minute per token, 600/hour per artifact, 120/hour per source address. See Limits.

Collection reads return ETag; send If-None-Match for an unchanged page's 304. Both responses use Cache-Control: private, no-cache. Authorization and budgets still apply. Cache config in your runner; share one fetch per workflow group. Refresh with jitter, never faster than config changes. Honour Retry-After seconds with backoff; keep last-known-good. Over-budget calls get 429; continued polling meets existing key/endpoint throttles.

Every MCP tool has an endpoint

MCP toolEndpoints
super_whoamiGET /v1/whoami
super_publishPOST /v1/artifacts
super_uploadPOST /v1/uploads
super_artifactsGET /v1/artifacts, GET /v1/artifacts/{artifact}, GET /v1/artifacts/{artifact}/file, POST /v1/artifacts/{artifact}/open
super_artifact_setPATCH /v1/artifacts/{artifact}, DELETE /v1/artifacts/{artifact}, POST /v1/artifacts/{artifact}/versions/{version}/serve, POST /v1/artifacts/{artifact}/live, DELETE /v1/artifacts/{artifact}/live
super_dataGET /v1/artifacts/{artifact}/data, GET /v1/data
super_data_queryPOST /v1/data/query
super_data_writePOST /v1/artifacts/{artifact}/data/rows
super_data_schemaGET /v1/schema, PUT /v1/schema
super_entriesGET /v1/collections/{collection}/entries, POST /v1/collections/{collection}/entries
super_secret_usePOST /v1/collections/{collection}/entries/{entry}/fields/{field}/use
super_templatesGET /v1/templates, GET /v1/templates/{slug}, GET /v1/templates/{slug}/history, GET /v1/templates/{slug}/feedback
super_template_savePOST /v1/templates
super_template_actionPOST /v1/templates/{slug}/actions
super_release_templatePOST /v1/templates/{slug}/release
super_projectGET /v1/projects, POST /v1/projects, GET /v1/projects/{project}, PATCH /v1/projects/{project}, DELETE /v1/projects/{project}, POST /v1/projects/{project}/artifacts, DELETE /v1/projects/{project}/artifacts/{remove}
super_sharePOST /v1/artifacts/{artifact}/shares, POST /v1/projects/{project}/shares, POST /v1/workspace/shares
super_commentsGET /v1/artifacts/{artifact}/comments, POST /v1/artifacts/{artifact}/comments/responses, GET /v1/artifacts/{artifact}/comments/screenshots/{screenshot}, GET /v1/mentions
super_signaturesGET /v1/signatures, POST /v1/signatures, GET /v1/signatures/{request}, POST /v1/signatures/{request}/remind, POST /v1/signatures/{request}/withdraw
super_helpGET /v1/help