generated: '2026-08-13'
method: searched
source: >-
  https://docs.segmentstream.com/.well-known/agent-skills/segmentstream/skill.md
  (provider-published, discovered via the A2A agent card) + docs.segmentstream.com/mcp/supported-tools
detail: >-
  SegmentStream publishes its own Agent Skill and advertises it from its A2A
  agent card, so the primary entry below is the provider's file saved verbatim
  rather than anything authored here. The three flow-specific skills generated
  in the previous round are retained beneath it: they decompose the provider's
  single broad skill into the marquee flows and name the exact MCP tools each
  one calls, which the provider's skill does not do.
skills:
  - file: segmentstream-provider-skill.md
    name: Segmentstream
    method: searched
    provenance: first-party
    source: https://docs.segmentstream.com/.well-known/agent-skills/segmentstream/skill.md
    discovered_via: a2a/segmentstream-agent-card.json (skills[0].url)
    api: mcp/segmentstream-mcp.yml
    fetched: '2026-08-13'
    http_status: 200
    content_type: text/markdown
    bytes: 15102
    description: >-
      Use when building marketing attribution models, analyzing campaign
      performance, optimizing budget allocation across channels, configuring
      event tracking and data sources, or querying attribution data through
      natural language via MCP.
    covers:
      - Product summary and when to reach for the skill
      - Core configuration workflow (8 steps)
      - Conversion types, attribution models, report dimensions and metrics
      - Decision guidance tables (warehouse, conversion type, model, MCP vs UI)
      - Four worked workflows including querying via MCP
      - Ten common gotchas and a ten-item verification checklist
    note: >-
      Saved verbatim. Frontmatter carries metadata.mintlify-proj: segmentstream
      and version "1.0". Written against the product surface rather than the tool
      surface — it names only a handful of MCP tools inline and does not
      enumerate operations, which is why the derived skills below are kept.
  - file: segmentstream-measure-and-optimize-spend.md
    name: Measure attribution and optimize ad spend
    method: generated
    provenance: derived
    api: mcp/segmentstream-mcp.yml
    operations: [list_active_projects, get_project, run_report, run_report_timeseries, list_portfolios, get_portfolio_optimization, get_portfolio_history]
  - file: segmentstream-run-geo-incrementality-test.md
    name: Read geo-incrementality experiment results
    method: generated
    provenance: derived
    api: mcp/segmentstream-mcp.yml
    operations: [list_active_projects, list_experiments, get_experiment, list_experiment_custom_parameter_keys]
  - file: segmentstream-configure-conversions.md
    name: Configure conversions and query performance
    method: generated
    provenance: derived
    api: mcp/segmentstream-mcp.yml
    operations: [list_active_projects, list_conversions, get_conversion, create_conversion, list_conversion_fields, get_conversion_statistics, get_conversions_by_country]
related:
  prompt_library:
    url: https://segmentstream.com/ai-skills
    kind: prompt-templates
    count: 30+
    note: >-
      SegmentStream also publishes a page it calls "AI Skills". These are
      copy-paste prompt templates for Claude, Codex and Cursor across six
      categories — not installable Agent Skill files, with no download URL,
      package or repository. Recorded here so the naming collision does not read
      as a second skill artifact.