generated: '2026-09-13'
method: searched
source: https://www.autoura.com/core/pai/skill.md
note: >-
  Autoura PUBLISHES its own agent skills. The homepage advertises them with a
  <link rel="ai-skill" href="/core/pai/skill.md" type="text/markdown"> element in
  the document head, and the skill document indexes a further thirteen markdown
  reference documents under /core/pai/docs/. Nothing in this directory was
  authored by API Evangelist -- every file is the provider's own document saved
  verbatim. The PlanMyVisit files are Autoura's consumer-facing agent guide for
  the SAME MCP endpoint (https://api.autoura.com/api/mcp); Autoura's own skill.md
  names PlanMyVisit as its product and links to that guide.
discovery:
  mechanism: link rel="ai-skill"
  html_head: '<link rel=ai-skill href=/core/pai/skill.md type=text/markdown>'
  observed_on: https://www.autoura.com/
  fetched: '2026-09-13'
skills:
  - file: autoura-agent-skill.md
    name: Autoura — AI Agent Integration
    description: >-
      The B2B operating guide for an agent working inside Autoura: what Autoura
      is, when to hand off to PlanMyVisit, how to authenticate, the three-tier
      guidance model (getting started / strategy / execution), the PAI (Personal
      AI) identity model, tasks as the operational unit, entity definitions with
      their ID prefixes, and the contact routes to a human.
    api: mcp/autoura-mcp.yml
    source: https://www.autoura.com/core/pai/skill.md
    method: searched
    http_status: 200
    content_type: text/markdown
    bytes: 9850
    operations: []
    operations_note: >-
      The skill deliberately names no fixed tool list. Autoura instructs agents to
      "discover available MCP tools dynamically at runtime rather than assuming
      tool names or schemas". brand_overview is the single tool it names by name.
  - file: autoura-planmyvisit-agent-skill.md
    name: PlanMyVisit — agent operating guide
    description: >-
      Autoura's consumer-side agent guide: authenticate over the email-code flow
      or WebMCP, discover tools, find or resume the one shared visit plan, load
      preferences, resolve the venue, then create/refine/resume. Carries the
      before-visit and during-visit workflows and the consent rules.
    api: mcp/autoura-mcp.yml
    source: https://www.planmyvisit.to/skill.md
    method: searched
    http_status: 200
    bytes: 9854
    operations:
      - visitplan_get
      - visitplans_search
      - visitplan_new
      - visitplan_update
      - preferences_get
      - visit_venues_search
      - visit_knowledge_get
  - file: autoura-planmyvisit-tool-reference.md
    name: PlanMyVisit selected tool reference
    description: >-
      The provider's own prose description of the MCP tool surface, grouped into
      venue/knowledge, preference, companion, visit-plan and feedback tools. The
      provider states explicitly that it is "a selected capability reference, not
      a complete catalogue or a fixed schema".
    api: mcp/autoura-mcp.yml
    source: https://www.planmyvisit.to/docs/ai/tools.md
    method: searched
    http_status: 200
    bytes: 11776
  - file: autoura-planmyvisit-authentication.md
    name: PlanMyVisit authentication reference
    description: >-
      The identity model (human owner + optional PAI), the three POST identity
      endpoints, the 48-hour access-token lifecycle, the WebMCP session model and
      the three email-access scenarios an agent may find itself in.
    api: authentication/autoura-authentication.yml
    source: https://www.planmyvisit.to/docs/ai/authentication.md
    method: searched
    http_status: 200
    bytes: 8385
supporting_documents:
  directory: skills/autoura-pai-docs/
  source_base: https://www.autoura.com/core/pai/docs/
  note: >-
    The thirteen reference documents the skill indexes, saved verbatim. Autoura
    splits them into Core reference, Getting started, Strategy (guidance only --
    "do not execute directly from these") and Execution (procedural), plus
    Operations. This three-way split is unusual and is itself the finding: the
    provider tells the agent which documents are safe to act on.
  files:
    - {file: authentication.md, tier: core-reference}
    - {file: getting_started.md, tier: getting-started}
    - {file: visit_strategy.md, tier: strategy}
    - {file: autoura_strategy.md, tier: strategy}
    - {file: visit_config.md, tier: execution}
    - {file: route_config.md, tier: execution}
    - {file: stop_config.md, tier: execution}
    - {file: facts.md, tier: execution}
    - {file: content_gap_summary_array.md, tier: execution}
    - {file: character_config.md, tier: execution}
    - {file: offer_config.md, tier: execution}
    - {file: script_config.md, tier: execution}
    - {file: tasks.md, tier: operations}