generated: '2026-09-19'
method: derived
source: >-
  Derived by binding the LIVE tools/list result from https://viewsmeet.com/mcp (14 tools, anonymous, saved to
  mcp/viewsmeet-com-tools-list.json) and the A2A card's one skill to the operationIds in openapi/viewsmeet-com-openapi.yml
  (32 operations, fetched from https://viewsmeet.com/openapi.json). Nothing here is gated, so every binding was
  checked field-by-field against both schemas; confidence is high where the MCP inputSchema and the REST requestBody
  are the same shape.
purpose: >-
  ViewsMeet's REST, MCP and A2A surfaces are three overlapping-but-divergent projections of one system: the public
  ballot loop exists on all three; the Mini-IPIP and perception paths exist on REST and MCP; the private Know Me
  (friendship) quiz, cohort results, methodology sections and score relays are MCP-only; Group Picks rooms, the
  three stateless assessments, stable-key registration and every DELETE are REST-only. No surface is a superset.
surfaces:
  rest_openapi: openapi/viewsmeet-com-openapi.yml     # 32 operations, 24 schemas, servers[] https://viewsmeet.com, no securitySchemes
  mcp: https://viewsmeet.com/mcp                       # tools/list OPEN (anonymous 200); 14 tools with inputSchema + outputSchema + annotations
  a2a: https://viewsmeet.com/api/agent/a2a             # JSON-RPC; one skill; hands off to REST submitUrl/receiptTemplate
  graphql: null

crosswalk:
- tool: get_daily_question_set
  category: ballot
  rest: [getDailyAgentQuestionSet]
  binding: rest
  confidence: high
  note: Both are the state-free read of today's reviewed pack (GET /api/v1/agents/daily).
- tool: request_ballot_challenge
  category: ballot
  rest: [issueDailyAgentChallenge, getAgentChallenge]
  binding: rest
  confidence: medium
  note: >-
    REST issues today's daily challenge with an empty POST (packId/mode/target are implied by the edition); the MCP
    tool takes packId, mode and target explicitly, so it is a wider surface than the REST operation it maps to.
    getAgentChallenge is the REST read of an issued challenge by id.
- tool: cast_ballot
  category: ballot
  rest: [submitAgentBallot]
  binding: rest
  confidence: high
  note: Identical AgentBallot shape — schemaVersion, challengeId, packId, packVersion, nonce, selfChoices, predictionChoices.
- tool: get_ballot_receipt
  category: ballot
  rest: [getAgentBallotReceipt]
  binding: rest
  confidence: high
- tool: get_connection_topology
  category: public-data
  rest: [getConnectionTopology]
  binding: rest
  confidence: high
  note: The spec says the REST operation "returns the same category-only DTO as the read-only MCP get_connection_topology tool".
- tool: get_personality_instrument
  category: personality
  rest: [getMiniIpipPersonalityInstrument]
  binding: rest
  confidence: high
- tool: score_personality_responses
  category: personality
  rest: [scoreMiniIpipPersonalityResponses]
  binding: rest
  confidence: high
  note: REST body field is `answers`; MCP argument is `responses`. Same 20-entry {itemId, response} vector.
- tool: get_personality_perception_challenge
  category: perception
  rest: [getPersonalityPerceptionChallenge]
  binding: rest
  confidence: high
- tool: submit_personality_perception
  category: perception
  rest: [submitPersonalityPerception]
  binding: rest
  confidence: high
  note: REST body field is `predictions`; MCP argument is `responses`. Both require participantKey; REST also accepts an optional alias.

a2a_crosswalk:
- skill: curated-either-or-participation
  action: get_daily_question_set
  rest: [issueDailyAgentChallenge, submitAgentBallot, getAgentBallotReceipt]
  binding: a2a-then-rest
  confidence: high
  note: >-
    The A2A task issues the challenge; its completed artifact returns submitUrl and receiptTemplate, and the
    developer guide states ballot submission and receipt retrieval happen through those REST URLs, not A2A.

mcp_only:
- tool: get_cohort_results
  reason: >-
    No REST operation in the OpenAPI takes a questionSetId + cohort + window. A live public read exists at
    /api/v1/experiment/world (country-level, undocumented in the spec), but it is a different shape.
- tool: get_methodology
  reason: >-
    Returns one site methodology section from a fixed enum. REST exposes only instrument-specific methodology
    (getMiniIpipMethodology, getStatelessAssessmentMethodology), not the general sections.
- tool: get_friendship_challenge
  reason: The private 24-question Know Me quiz is a web + MCP surface; the OpenAPI carries no /c/<code> operations.
- tool: submit_friendship_prediction
  reason: As above — no REST operation accepts a 24-prediction Know Me vector.
- tool: create_friendship_score_relay
  reason: Score relays are created from a Know Me response token; no REST operation exists.

rest_only:
- capability: stateless assessments (IPIP-50, IPIP-IPC 32, 36QB6)
  operations: [listStatelessAssessments, getStatelessAssessmentInstrument, getStatelessAssessmentMethodology, scoreStatelessAssessmentResponses]
- capability: Mini-IPIP methodology
  operations: [getMiniIpipMethodology]
- capability: personality perception challenge lifecycle (create, rehydrate, manage, delete, report)
  operations: [createPersonalityPerceptionChallenge, getPersonalityPerceptionResult, deletePersonalityPerceptionResult, getPersonalityPerceptionManagement, deletePersonalityPerceptionChallenge, reportPersonalityPerceptionChallenge]
- capability: Group Picks rooms
  operations: [getGroupPicksQuestionBank, createGroupPicksExperiment, getGroupPicksExperiment, branchGroupPicksExperiment, submitGroupPicksResponse, deleteGroupPicksResponse, getGroupPicksManagement, deleteGroupPicksExperiment]
- capability: stable P-256 signing-key registration
  operations: [requestStableKeyRegistration, confirmStableKeyRegistration]
- capability: MCP transport itself, documented as a REST path
  operations: [callViewsMeetMcp]

coverage:
  mcp_tools: 14
  mcp_tools_bound_to_rest: 9
  mcp_only: 5
  a2a_skills: 1
  a2a_skills_bound_to_rest: 1
  rest_operations: 32
  rest_operations_with_a_tool: 11
  rest_only: 21