Scale ai for AI agents
Scale AI provides data labeling and annotation services, letting teams create and manage tasks like image, video, and text annotation to train machine learning models. Connect it once through Arc0, and your agent, or Claude, ChatGPT and Cursor, can use it through one MCP endpoint, limited to what each user approved.
What agents do in Scale ai.
Create an annotation task
Sets up a new image or video annotation task with the label instructions a model needs.
Check a batch's status
Looks up whether a batch of annotation tasks is finished before pulling results for training.
Finalize a batch only after review
Locks a batch of completed annotations only once someone confirms the labels are ready to ship.
40 Scale ai actions, graded by risk.
Every Scale ai action is tagged read, write or destructive, so one policy covers the whole app and new actions inherit the right default.
read
14Look things up. Allowed by default.
- scale_ai.get_batchGet Batch
- scale_ai.get_teamsGet Teams
- scale_ai.get_assetsGet Assets
- scale_ai.list_tasksList Tasks
- scale_ai.get_projectGet Project
- scale_ai.get_v1_taskGet Task by ID
- scale_ai.list_batchesList Batches
- scale_ai.list_projectsList Projects
- scale_ai.get_batch_statusGet Batch Status
- scale_ai.get_fixless_auditsGet Fixless Audits
- scale_ai.get_studio_batchesGet Studio Batches
- scale_ai.get_quality_labelersGet Quality Labelers
- scale_ai.get_studio_assignmentsGet Studio Assignments
- scale_ai.get_v1_task_response_urlGet Secure Task Response URL
write
22Create and change things. Allow, or ask the user first.
- scale_ai.create_batchCreate Batch
- scale_ai.add_task_tagsAdd Task Tags
- scale_ai.update_task_unique_idUpdate Task Unique ID
- scale_ai.add_studio_assignmentsAdd Studio Assignments
- scale_ai.post_v1_task_send_callbackRe-send Task Callback
- scale_ai.create_text_collection_taskCreate Text Collection Task
- scale_ai.create_image_annotation_taskCreate Image Annotation Task
- scale_ai.create_lidar_annotation_taskCreate Lidar Annotation Task
- scale_ai.create_video_annotation_taskCreate Video Annotation Task
- scale_ai.create_lidar_segmentation_taskCreate LiDAR Segmentation Task
- scale_ai.create_document_transcription_taskCreate Document Transcription Task
- scale_ai.create_segmentation_annotation_taskCreate Segmentation Annotation Task
- scale_ai.create_named_entity_recognition_taskCreate Named Entity Recognition Task
- scale_ai.create_video_playback_annotation_taskCreate Video Playback Annotation Task
- scale_ai.import_fileImport File
- scale_ai.upload_fileUpload File
destructive
4Delete, cancel or archive. Ask first, or deny outright.
- scale_ai.delete_task_tagsDelete Task Tags
- scale_ai.delete_task_unique_idDelete Task Unique ID
- scale_ai.remove_studio_assignmentsRemove Studio Assignments
- scale_ai.reset_batch_prioritiesReset Batch Priorities
Scale ai in three steps.
- 01Your users connect Scale aiThey add their Scale ai api key on Arc0 Connect, under your brand. It goes straight into the vault.
- 02You set the rulesReads run, writes like “create Batch” can wait for the user, and “delete Task Tags” can be denied outright.
- 03Any agent can actYour agent calls Scale ai through the Arc0 SDK or MCP, and so can Claude, ChatGPT and Cursor. Every call lands on the audit log.
await arc0.policies.set('scale_ai', { read: 'allow', write: 'ask', // create_batch destructive: 'deny', // delete_task_tags }) # Claude Code: the same connection, one URL $ claude mcp add --transport http arc0 \ https://mcp.arc0.ai/u/u_8f2
How Scale ai connects.
Users add their Scale ai api key on Arc0 Connect. It is encrypted in the vault, never shown to the model, and each user can rotate or revoke it at any time.
The same Scale ai connection serves your agent over MCP and your own backend over REST and the proxy, so a user connects once. How Arc0 handles credentials →
- AUTH
- API key
- CREDENTIALS
- Per-tenant encrypted vault
- MODEL SEES
- Results only, never credentials
- AUDIT LOG
- Every call, on every plan
Use Scale ai from any agent.
ClaudeChatGPTCursor
Codex
VS CodeOpenAI Agents SDKClaude Agent SDK
Vercel AI SDK
Mastra
LangGraph
Scale ai and Arc0, answered.
Can I use Scale ai with Claude, ChatGPT or Cursor?
Yes. Connect Scale ai to Arc0 once, then add your Arc0 MCP URL to Claude, ChatGPT, Cursor, Claude Code or any other remote-MCP client. Each assistant only gets the Scale ai actions you allow.
How do users connect Scale ai?
Users add their Scale ai api key on Arc0 Connect. It is encrypted in the vault, never shown to the model, and each user can rotate or revoke it at any time.
Which Scale ai actions can my agent take?
40 in total: 14 read, 22 write and 4 destructive, such as “create Batch”. Your policies decide which of them each agent may call.
Can I stop my agent from deleting things in Scale ai?
Yes. Actions like “delete Task Tags” are graded destructive. Set destructive actions to deny, or to ask so the user approves each one, and blocked calls still show up on the audit log.
Can my own backend call Scale ai too?
Yes. The same Scale ai connection is available over REST and through the Arc0 proxy, so your product and your agent share one connection per user.
Plug Scale ai into your agent.
Your users connect Scale ai once, under your brand. Your agent gets 40 actions behind your policies, with every call on the record.
Free to build · MCP + REST · Audit log on every plan