Data Agents API

Data Agents API is the governed data module behind dataset versions, manifests, execution jobs, lineage, decisions, insights, recommendations, and knowledge artifacts.

This is different from Data & Insights in the Slab5 console. Data & Insights is the user-facing analytics workspace for Object Analytics, Insight Builder, Insight Views, Custom BI, Analytics Sites, and runs. Data Agents API is the lower-level module that developers and agents use when they need durable data-governance records and execution lifecycle objects.

What it does

Use Data Agents API when an analytics workflow needs traceability:

• Create dataset versions and manifests that describe the source data and shape.
• Track lineage edges between source objects, datasets, artifacts, insights, and recommendations.
• Create decisions that a human can approve before an insight or recommendation is used.
• Execute jobs and inspect their status, timing, attempts, output, and review state.
• Publish knowledge artifacts for retrieval by Slab5 Guide or other agent workflows.

Objects

Object	Purpose
execution_planes	Named execution targets and policy settings for data-agent jobs.
dataset_versions	Versioned dataset definitions with schema, freshness, and source references.
dataset_manifests	Structured manifests that explain dataset fields and lineage.
execution_jobs	Run records for data-agent execution and review.
decisions	Approval records for generated datasets, artifacts, insights, or recommendations.
insights	Published findings tied to source objects and governance context.
recommendations	Suggested actions that can be reviewed, accepted, or completed.
knowledge_artifacts	Markdown, JSON, or other retrieval-ready artifacts produced by governed workflows.

REST endpoints

Data Agents API routes use the /v1/data-agents/... namespace:

• GET /v1/data-agents/execution-planes
• POST /v1/data-agents/execution-planes
• PATCH /v1/data-agents/execution-planes/{execution_plane_id}
• GET /v1/data-agents/dataset-versions
• POST /v1/data-agents/dataset-versions
• GET /v1/data-agents/manifests
• POST /v1/data-agents/manifests
• GET /v1/data-agents/execution-jobs
• POST /v1/data-agents/execution-jobs
• GET /v1/data-agents/execution-jobs/{job_id}
• PATCH /v1/data-agents/execution-jobs/{job_id}
• GET /v1/data-agents/lineage
• POST /v1/data-agents/lineage
• GET /v1/data-agents/decisions
• POST /v1/data-agents/decisions
• PATCH /v1/data-agents/decisions/{decision_id}
• GET /v1/data-agents/insights
• POST /v1/data-agents/insights
• GET /v1/data-agents/recommendations
• POST /v1/data-agents/recommendations
• PATCH /v1/data-agents/recommendations/{recommendation_id}
• GET /v1/data-agents/source-objects
• POST /v1/data-agents/source-objects
• GET /v1/data-agents/knowledge-artifacts
• POST /v1/data-agents/knowledge-artifacts

MCP tools

Agents can use Data Agents tools for the same lifecycle:

• propose_data_agent_dataset
• approve_data_agent_decision
• publish_data_agent_insight
• execute_data_agent_recommendation
• review_data_agent_execution_job
• create_content_performance_knowledgebase

See the MCP tools reference for the full list of create, list, update, dispatch, and review tools.

Permissions

Use data_agents:read for listing and inspection, data_agents:write for creating governed objects, data_agents:execute for dispatching jobs or recommendations, and data_agents:admin for execution-plane administration.

Data Agents API workflows often also need bi:read or bi:write when they use Data & Insights objects, plus module-specific scopes for source records such as CRM, Support, CMS, or Assets.