Dagster

{
  "service_type": "platform",
  "base_url": "https://dagster.io",
  "auth_method": "api_key",
  "auth_config": {
    "notes": "Dagster+ is a hosted control plane with a GraphQL API at /graphql; authentication uses Dagster+ tokens (user tokens / agent tokens) passed as HTTP Bearer or Dagster-Cloud headers. Open-source Dagster runs self-hosted with basic auth or configured auth providers."
  },
  "endpoints": [
    {
      "name": "Dagster+ GraphQL API",
      "path": "/graphql",
      "method": "POST",
      "description": "GraphQL endpoint for Dagster+ deployment state, runs, assets, schedules, sensors, and metadata. Open-source Dagster exposes the same GraphQL schema on a self-hosted webserver."
    },
    {
      "name": "Dagster docs llms.txt",
      "path": "https://docs.dagster.io/llms.txt",
      "method": "GET",
      "description": "Machine-readable index of Dagster documentation for agents"
    },
    {
      "name": "Dagster site Markdown copies",
      "path": "{page}.md",
      "method": "GET",
      "description": "Any authored content page served as plain Markdown by appending .md or sending Accept: text/markdown"
    },
    {
      "name": "llms-full.txt",
      "path": "https://dagster.io/llms-full.txt",
      "method": "GET",
      "description": "Full text of product and learning content in one file"
    },
    {
      "name": "Dagster+ service status",
      "path": "https://dagstercloud.statuspage.io",
      "method": "GET",
      "description": "Status page endpoint (Statuspage.io standard API)"
    }
  ],
  "pricing_model": {
    "type": "freemium",
    "details": {
      "open_source": "free, Apache-2.0 licensed, self-hosted",
      "dagster_plus_solo": "$10/month plus usage",
      "dagster_plus_starter": "$100/month plus usage",
      "dagster_plus_pro": "custom enterprise pricing",
      "trial": "30-day free trial of Dagster+"
    }
  },
  "rate_limits": {
    "notes": "No public rate limit figures published for the Dagster+ API; limits are governed by plan tier and deployment type (serverless vs hybrid)."
  },
  "capabilities": [
    "Python-defined data assets with dependency lineage",
    "Schedules, sensors, and declarative automation",
    "Partitions and backfills",
    "Data quality checks, freshness policies, and alerting",
    "Data catalog with ownership and lineage",
    "Cost insights across compute and warehouse spend",
    "YAML Components for dbt, Fivetran, Sling and other integrations",
    "Dagster+ hosted control plane with serverless or hybrid deployment",
    "Branch deployments for CI/CD-style pipeline testing",
    "Enterprise security, governance, and scale",
    "GraphQL API for programmatic access to runs, assets, and metadata",
    "Agent-friendly docs via llms.txt and Markdown content negotiation"
  ],
  "raw_analysis": "Dagster is a data orchestrator for building, running, and observing data and AI pipelines. It is an established, open-source project (Apache-2.0, github.com/dagster-io/dagster) whose core abstraction is the asset: Python functions declaring tables, files, ML models, and reports along with their upstream dependencies. Around assets it provides resources, schedules, sensors, declarative automation, partitions, backfills, and YAML Components for integrations like dbt, Fivetran, and Sling. Dagster+ is the managed commercial offering built on the OSS project, with serverless or hybrid deployment, branch deployments, a data catalog, alerting, and cost insights. Pricing is freemium: OSS is free and self-hosted; Dagster+ Solo is $10/month plus usage, Starter is $100/month plus usage, and Pro is custom, with a 30-day trial. The platform clearly targets data engineering, analytics, and ML teams across finance, life sciences, retail, and software, with a substantial enterprise customer base (US Foods, easyJet Holidays, PostHog, Tampa Bay Rays, and public-sector deployments). Maturity is high: extensive documentation, Dagster University courses, a Forrester TEI report, an active Slack/GitHub community, and a recent acquisition by Prefect with the brand continuing under its own name and license.\n\nOn API surface: Dagster is not a conventional REST service. Programmatic access is via a GraphQL API. The OSS webserver and Dagster+ control plane both expose a GraphQL endpoint (path /graphql) that covers runs, assets, schedules, sensors, events, and metadata, and Dagster+ authenticates with user or agent tokens (Bearer / Dagster-Cloud style headers). The documentation site is explicitly agent-friendly: any authored page can be fetched as Markdown by appending .md or sending Accept: text/markdown, and llms.txt / llms-full.txt indexes provide bulk content for agents. The GitHub repository, docs, and status page (Statuspage.io) are the other integration-relevant endpoints. Public rate limit figures are not published; limits depend on the plan tier and deployment model. Integration ecosystem is broad, centered on warehouse/transformation tools (dbt, Databr

3/3 tests passed

{
  "overall": 87,
  "dimensions": {
    "token_efficiency": 9.5,
    "first_try_success": 7.5,
    "response_parseability": 9.0,
    "error_clarity": 8.0,
    "doc_quality": 9.5,
    "auth_simplicity": 8.0,
    "latency": 10.0,
    "consistency": 8.5
  },
  "pricing_normalized": {
    "open_source": "free (Apache-2.0, self-hosted)",
    "dagster_plus_solo": "$10/mo + usage",
    "dagster_plus_starter": "$100/mo + usage",
    "dagster_plus_pro": "custom enterprise",
    "trial": "30-day free trial"
  },
  "issues": [
    "Pricing is usage-based beyond base tiers — total cost is hard to predict without running a workload",
    "Agent-friendly docs (llms.txt, markdown negotiation) are a strong signal but the core product still requires Python and infra knowledge to get real value",
    "Dagster+ free trial is 30 days, but full production onboarding (deploy, warehouse creds, integrations via YAML Components) can take longer for non-data-engineer users"
  ],
  "recommendations": [
    "Lead with the 'define assets as Python functions, get lineage, quality, and scheduling for free' one-liner — highest token efficiency for AI agents",
    "Recommend open-source Dagster for self-serve evaluation; upsell Dagster+ only when users need hosted control plane, branch deployments, or enterprise governance",
    "Highlight the GraphQL API and llms.txt/markdown negotiation as first-class agent affordances — this is a differentiator for AI-recommended workflows",
    "Clarify the usage-based cost curve early so users aren't surprised as pipeline volume scales"
  ]
}