{"schema":"https://ver.cy/schemas/card/1.0.0","id":"vr.wm-ai-017","code":"wm-ai-017-tool-description","url":"https://ver.cy/models/wm-ai-017-tool-description/","name":"Tool Description","alternateNames":[],"kind":"world-model","status":"published","version":"0.1.0","language":"en","classifiers":{"family":"World Models","category":"Information and virtual systems","entryKind":"entity","plane":"","domain":["INF.AI.TOL"],"industry":["Cross-industry"],"navPath":"NAV.INF.AI.TOL","tags":["tool","description","inf.ai.tol"],"facets":{}},"whatItIs":"One governed tool-description entity with separately identified revisions and optional implementation bindings. The root describes a callable capability, not an agent, invocation, server, physical tool or permission grant. The hierarchy and operations below are proposed local design, informed by selected open specifications, not their universal requirements.","purpose":"Describe one identifiable callable tool contract so an agent can assess applicability, required inputs, declared effects and authority dependencies before referring execution elsewhere.","scope":{"in":["Description identity, purpose, callable selection and bounded use cases.","Input and result contract declarations, dialects, examples, protocol bindings and error semantics.","Declared effects, uncertainty, risk, authority prerequisites, policy references and review evidence.","Description revision, provenance, discovery observations, retirement and conceptual interoperability mappings."],"out":["Executable tool implementation, live invocation, scheduler, transport session, task execution or external side effects.","Credential issuance or storage, permission enforcement, consent decisions and authoritative access grants.","Agent instruction hierarchy, agent lifecycle, configuration ownership and handoff message semantics.","Runtime result records, audit trail masters, evaluator logic, incident handling and domain-specific safety certification.","Physical equipment identity or measurements; operational telemetry remains externally mastered."],"boundaries":[{"neighbor":"WM-AI-005 Prompt / Agent Configuration","distinction":"Registry parent is a candidate contextual relationship, not accepted inheritance. A configuration may reference this description revision; configuration policy and instruction precedence stay outside this entity. No explicit relation-ledger rows exist for this model."},{"neighbor":"Tool implementation and service binding","distinction":"A description can have multiple reviewed bindings. Endpoint and operation references do not transfer implementation ownership or establish availability."},{"neighbor":"Invocation, result and execution evidence","distinction":"Contract and failure categories are local; individual arguments, outputs, execution state and side effects belong to external records."},{"neighbor":"Authorization and policy authority","distinction":"Required scopes, review gates and policy references describe prerequisites. They cannot mint permission, authenticate a caller or authorize execution."},{"neighbor":"Schema resources and provenance records","distinction":"Pin external schemas and evidence by identity and version. Local copies are projections with attribution, not new authoritative masters."}]},"distinguishingFeatures":["Unlike WM-AI-005 Prompt / Agent Configuration: Registry parent is a candidate contextual relationship, not accepted inheritance. A configuration may reference this description revision; configuration policy and instruction precedence stay outside this entity. No explicit relation-ledger rows exist for this model.","Unlike Tool implementation and service binding: A description can have multiple reviewed bindings. Endpoint and operation references do not transfer implementation ownership or establish availability.","Unlike Invocation, result and execution evidence: Contract and failure categories are local; individual arguments, outputs, execution state and side effects belong to external records.","Unlike Authorization and policy authority: Required scopes, review gates and policy references describe prerequisites. They cannot mint permission, authenticate a caller or authorize execution.","Unlike Schema resources and provenance records: Pin external schemas and evidence by identity and version. Local copies are projections with attribution, not new authoritative masters."],"structure":{"bundles":[{"id":"bundle-identity","name":"Identity and purpose","description":"Proposed local grouping for identity and purpose.","layers":[{"id":"layer-identity-scope","name":"Identity scope","description":"Context for scoped description identity within the description boundary.","findings":[{"id":"descriptor-identity","name":"Scoped description identity","description":"Keep the governed description ID, revision ID, protocol-local callable name and implementation reference separate. A display label is not a global identifier.","questions":[{"text":"Which master identifier and namespace identify this tool description independently of its callable name?","id":"descriptor-identity-q1","kind":"identity"},{"text":"What operation category and subject domain describe the tool without implying implementation inheritance?","id":"descriptor-identity-q2","kind":"classification"},{"text":"Which implementation or service references realize this description revision, if any?","id":"descriptor-identity-q3","kind":"relationship"},{"text":"Which accountable role maintains the description and can approve identity changes?","id":"descriptor-identity-q4","kind":"ownership"}]}]},{"id":"layer-purpose-scope","name":"Purpose and selection","description":"Context for applicability and limitations within the description boundary.","findings":[{"id":"selection-context","name":"Applicability and limitations","description":"State the intended outcome, suitable requests and excluded uses. Selection guidance is descriptive content; it has no authority to override governing instructions.","questions":[{"text":"What user outcome does this callable capability support?","id":"selection-context-q1","kind":"definition"},{"text":"Which request classes are outside the documented use boundary?","id":"selection-context-q2","kind":"constraint"},{"text":"Which conditions distinguish this tool from another tool with a similar name?","id":"selection-context-q3","kind":"decision"},{"text":"Which sanitized examples demonstrate appropriate and inappropriate selection?","id":"selection-context-q4","kind":"evidence"}]}]}]},{"id":"bundle-contract","name":"Input and result contracts","description":"Proposed local grouping for input and result contracts.","layers":[{"id":"layer-input-contract","name":"Input declaration","description":"Context for typed arguments and schema interpretation within the description boundary.","findings":[{"id":"argument-contract","name":"Typed arguments and schema interpretation","description":"Bind input constraints to an identified dialect and schema resource. Distinguish absent, null, default annotations, assertions and parameter serialization. Unknown vocabulary support is an explicit gap.","questions":[{"text":"Which input schema, dialect and required vocabularies govern argument interpretation?","id":"argument-contract-q1","kind":"requirement"},{"text":"How are omitted, null and defaulted values distinguished for each parameter?","id":"argument-contract-q2","kind":"constraint"},{"text":"How are parameter location, encoding and units represented in each supported binding?","id":"argument-contract-q3","kind":"interoperability"},{"text":"Which input cases test required fields, unknown properties and boundary values?","id":"argument-contract-q4","kind":"validation"}]}]},{"id":"layer-result-contract","name":"Result declaration","description":"Context for success and failure contract within the description boundary.","findings":[{"id":"result-semantics","name":"Success and failure contract","description":"Describe successful payload forms and error channels separately. Missing output schema means no declared structural guarantee. Schema validity is not evidence that an operation succeeded or returned true information.","questions":[{"text":"Which structured fields, content forms and resource references can a successful result contain?","id":"result-semantics-q1","kind":"composition"},{"text":"How can a consumer distinguish protocol failure, tool failure, partial result and successful completion?","id":"result-semantics-q2","kind":"exception"},{"text":"What semantic evidence is needed beyond a schema-valid result?","id":"result-semantics-q3","kind":"quality"},{"text":"Which result fields or examples require redaction or restricted disclosure?","id":"result-semantics-q4","kind":"privacy"}]}]}]},{"id":"bundle-binding","name":"Bindings and discovery","description":"Proposed local grouping for bindings and discovery.","layers":[{"id":"layer-protocol-binding","name":"Protocol binding","description":"Context for explicit operation mapping within the description boundary.","findings":[{"id":"callable-binding","name":"Explicit operation mapping","description":"Map the contract to a transport-specific operation without conflating a server with one tool. Preserve relative-reference bases, media types and context needed to interpret names. Binding remains optional for an abstract draft.","questions":[{"text":"Which protocol version and operation locator identify each supported callable binding?","id":"callable-binding-q1","kind":"interoperability"},{"text":"Which schema and security references must resolve within the binding context?","id":"callable-binding-q2","kind":"relationship"},{"text":"Which declared execution location or data-residency policy constrains a binding, if applicable?","id":"callable-binding-q3","kind":"spatial"},{"text":"Which mapping losses or unsupported extensions prevent interchangeable use of two bindings?","id":"callable-binding-q4","kind":"validation"}]}]},{"id":"layer-discovery-freshness","name":"Discovery observation","description":"Context for observed discovery and freshness within the description boundary.","findings":[{"id":"availability-context","name":"Observed discovery and freshness","description":"Keep listing observations separate from description identity and runtime availability. Discovery may be paginated and context dependent; absence in one view does not prove deletion. Recheck changed material before use under the adopting policy.","questions":[{"text":"Which source and observer produced the tool listing used for this description?","id":"availability-context-q1","kind":"provenance"},{"text":"When was the listing observed and when does local policy require revalidation?","id":"availability-context-q2","kind":"temporal"},{"text":"Which caller or visibility context limits interpretation of the observed tool list?","id":"availability-context-q3","kind":"access"},{"text":"What evidence links a list-change notice to a reviewed description revision?","id":"availability-context-q4","kind":"event"}]}]}]},{"id":"bundle-behavior","name":"Behavior and operational limits","description":"Proposed local grouping for behavior and operational limits.","layers":[{"id":"layer-effect-declaration","name":"Declared effects","description":"Context for effects, reversibility and trust within the description boundary.","findings":[{"id":"effect-contract","name":"Effects, reversibility and trust","description":"Record claimed reads, writes, external contacts and reversibility with evidence status. Safe, read-only and idempotent concepts depend on the selected profile and are not interchangeable. Unknown effects remain unknown.","questions":[{"text":"Which external state is declared readable or mutable by the operation?","id":"effect-contract-q1","kind":"state"},{"text":"Under which profile and assumptions is a read-only, safe or idempotent claim made?","id":"effect-contract-q2","kind":"constraint"},{"text":"What review or test evidence supports each effect claim for the bound implementation?","id":"effect-contract-q3","kind":"evidence"},{"text":"Which effects are irreversible or require a separately authorized compensating operation?","id":"effect-contract-q4","kind":"exception"}]}]},{"id":"layer-execution-constraints","name":"Execution constraints","description":"Context for timing, retries and uncertain completion within the description boundary.","findings":[{"id":"operational-contract","name":"Timing, retries and uncertain completion","description":"Describe declared limits and recovery expectations without implementing execution. A transport failure or cancellation request does not prove rollback. Retry decisions require the binding and operation semantics, not a label alone.","questions":[{"text":"Is completion immediate or does the binding expose a separate status or cancellation operation?","id":"operational-contract-q1","kind":"process"},{"text":"What declared resource, payload or latency limits apply and what evidence qualifies them?","id":"operational-contract-q2","kind":"measurement"},{"text":"What must a caller do when a timeout leaves the operation outcome unknown?","id":"operational-contract-q3","kind":"exception"},{"text":"Under what conditions may a request be retried without unacceptable duplicate effects?","id":"operational-contract-q4","kind":"constraint"}]}]}]},{"id":"bundle-authority","name":"Authority and risk declarations","description":"Proposed local grouping for authority and risk declarations.","layers":[{"id":"layer-permission-prerequisites","name":"Permission prerequisites","description":"Context for required authority without grants within the description boundary.","findings":[{"id":"authority-contract","name":"Required authority without grants","description":"Reference identity, scopes, resource audience and approval policies without embedding secrets or granting permission. Authentication and transport authorization do not settle all business permissions. Non-HTTP bindings require their own authority profile.","questions":[{"text":"Which principal, resource audience and scope requirements apply to this binding?","id":"authority-contract-q1","kind":"authority"},{"text":"Which policy requires human approval or delegated authority before a material effect?","id":"authority-contract-q2","kind":"requirement"},{"text":"How are credential references separated from the publicly readable description?","id":"authority-contract-q3","kind":"security"},{"text":"Which missing, expired or insufficient authority conditions require refusal?","id":"authority-contract-q4","kind":"exception"}]}]},{"id":"layer-trust-and-risk","name":"Trust and risk review","description":"Context for untrusted descriptions and bounded use within the description boundary.","findings":[{"id":"risk-context","name":"Untrusted descriptions and bounded use","description":"Treat imported prose, annotations and schema references as untrusted data. Record hazard categories, data disclosure paths and required review at policy level. Provenance and signatures, if supplied, identify evidence rather than prove harmless behavior.","questions":[{"text":"Which trust boundary prevents tool prose or returned references from becoming governing instructions?","id":"risk-context-q1","kind":"security"},{"text":"Which data categories may leave the local context and under which declared purpose?","id":"risk-context-q2","kind":"privacy"},{"text":"Which policy limits schema resolution, external references and validation resource consumption?","id":"risk-context-q3","kind":"constraint"},{"text":"Which risk review applies to this description revision and what hazards remain unresolved?","id":"risk-context-q4","kind":"evidence"}]}]}]},{"id":"bundle-assurance","name":"Revision and assurance","description":"Proposed local grouping for revision and assurance.","layers":[{"id":"layer-revision-lifecycle","name":"Revision lifecycle","description":"Context for version, change and retirement within the description boundary.","findings":[{"id":"revision-contract","name":"Version, change and retirement","description":"Preserve revision provenance and effective applicability. A rename, schema change, broader authority or changed effect claim can invalidate a prior review. Retiring a description does not remove external implementations or historical execution records.","questions":[{"text":"What state marks this description as draft, reviewed, deprecated or retired?","id":"revision-contract-q1","kind":"lifecycle"},{"text":"Which effective interval and observation time qualify the revision?","id":"revision-contract-q2","kind":"temporal"},{"text":"Which prior revision and change rationale explain the current contract?","id":"revision-contract-q3","kind":"provenance"},{"text":"Which retention rule governs retired description payloads and minimal reference tombstones?","id":"revision-contract-q4","kind":"retention"}]}]},{"id":"layer-conformance-evidence","name":"Conformance evidence","description":"Context for evidence scope and adoption readiness within the description boundary.","findings":[{"id":"verification-context","name":"Evidence scope and adoption readiness","description":"Keep structural validation, binding compatibility, observed behavior and independent review as distinct claims. The present research supplies candidate fields and conceptual mappings, not an executable tool schema or conformance certificate.","questions":[{"text":"Which validator profile and fixtures support a structural-conformance claim?","id":"verification-context-q1","kind":"validation"},{"text":"What coverage and limitations qualify the contract test evidence?","id":"verification-context-q2","kind":"quality"},{"text":"Which field mappings to selected open specifications are lossless, lossy or not applicable?","id":"verification-context-q3","kind":"interoperability"},{"text":"Who may approve adoption after unresolved holds and independent review are considered?","id":"verification-context-q4","kind":"authority"}]}]}]}]},"agentConduct":{"may":["Register a local description: Proposed and unimplemented. Create a draft with governed identity, provenance and an explicit scope.","Validate a declared contract: Proposed and unimplemented. Proposed local validation against a pinned, supported profile; report schema and reference failures without executing the described tool.","Record binding review: Proposed and unimplemented. Record an approved mapping between a description revision and externally mastered callable operation.","Record policy review: Proposed and unimplemented. Attach a qualified reviewer assessment of declared effects and authority dependencies.","Propose a revision: Proposed and unimplemented. Compare prior and proposed declaration fields and record review invalidation triggers.","Retire a local description: Proposed and unimplemented. Mark a description deprecated or retired under the adopting retention policy."],"mustNot":["External masters own disposal of implementations, credentials, invocation records and evidence; this description cannot delete them.","Deny changes by default.","Grant purpose-specific read views and separate maintainer/reviewer authority; visibility never implies invocation permission."],"requiresHuman":[]},"ethics":{"considerations":["Sector-specific safety, privacy, approval and residency policies require qualified adoption review.","Credential issuance or storage, permission enforcement, consent decisions and authoritative access grants."],"affectedParties":[]},"owners":{"steward":"Accountable description maintainer role and authorized reviewer role, without vendor ownership assumptions.","roles":[{"name":"Description maintainer","responsibilities":["Maintain scoped contract, provenance and change proposals."]},{"name":"Contract reviewer","responsibilities":["Review schema interpretation, examples, mapping losses and declared effects."]},{"name":"Policy authority","responsibilities":["Define use constraints, approval prerequisites and retention rules without delegating grants to metadata."]},{"name":"Consumer agent","responsibilities":["Read permitted descriptions as data, check revision and invoke only through separately authorized execution controls."]},{"name":"Evidence custodian","responsibilities":["Maintain scoped evidence references and apply access and disposition decisions."]}],"masterSystems":[]},"relations":[{"target":"WM-AI-005","type":"references","note":"Candidate parent reconciliation: an agent configuration can select this description revision; configuration instructions and lifecycle remain external. No inheritance or required dependency is asserted."},{"target":"OpenAPI 3.1.1 Operation Object","type":"aligned","note":"Optional conceptual mapping for HTTP operation bindings; pin operation context and report losses. No universal equivalence to a tool is asserted."},{"target":"MCP 2025-06-18 Tool","type":"aligned","note":"Optional declaration projection with explicit protocol version, schema and annotation trust limits."},{"target":"W3C WoT Thing Description 1.1 ActionAffordance","type":"aligned","note":"Optional action/form mapping; this model does not become the Thing master."},{"target":"JSON Schema Draft 2020-12","type":"references","note":"Optional pinned schema resources and explicit vocabulary support; no automatic remote reference retrieval."},{"target":"W3C PROV-O","type":"aligned","note":"Conceptual attribution and revision relations for description evidence, not evidence truth or authorization."},{"target":"WM-AI-005 Prompt / Agent Configuration","type":"neighbor","note":"Registry parent is a candidate contextual relationship, not accepted inheritance. A configuration may reference this description revision; configuration policy and instruction precedence stay outside this entity. No explicit relation-ledger rows exist for this model."},{"target":"Tool implementation and service binding","type":"neighbor","note":"A description can have multiple reviewed bindings. Endpoint and operation references do not transfer implementation ownership or establish availability."},{"target":"Invocation, result and execution evidence","type":"neighbor","note":"Contract and failure categories are local; individual arguments, outputs, execution state and side effects belong to external records."},{"target":"Authorization and policy authority","type":"neighbor","note":"Required scopes, review gates and policy references describe prerequisites. They cannot mint permission, authenticate a caller or authorize execution."},{"target":"Schema resources and provenance records","type":"neighbor","note":"Pin external schemas and evidence by identity and version. Local copies are projections with attribution, not new authoritative masters."},{"target":"WM-AI-005","type":"parent"}],"interaction":{"identity":{"applicability":"required","items":["Authoritative master-system identifier with namespace and revision binding.","Governed global identifier or IRI.","UUID or ULID assigned by the adopting Dimension."]},"properties":{"applicability":"not-applicable","items":[]},"recognition":{"applicability":"optional","items":[]},"capabilities":{"applicability":"required","items":["Register a local description: Proposed and unimplemented. Create a draft with governed identity, provenance and an explicit scope.","Validate a declared contract: Proposed and unimplemented. Proposed local validation against a pinned, supported profile; report schema and reference failures without executing the described tool.","Record binding review: Proposed and unimplemented. Record an approved mapping between a description revision and externally mastered callable operation.","Record policy review: Proposed and unimplemented. Attach a qualified reviewer assessment of declared effects and authority dependencies.","Propose a revision: Proposed and unimplemented. Compare prior and proposed declaration fields and record review invalidation triggers.","Retire a local description: Proposed and unimplemented. Mark a description deprecated or retired under the adopting retention policy."]},"hazards":{"applicability":"optional","items":[]},"interfaces":{"applicability":"required","items":["RFC 9110: HTTP Semantics","PROV-O: The PROV Ontology"]},"context":{"applicability":"required","items":["No jurisdiction or commercial implementation is normative. The adopting authority must supply applicable policy and retention rules.","Selected MCP and OpenAPI versions are intentional research pins, not claims about the latest version."]}},"sources":[{"title":"MCP Tools","url":"https://modelcontextprotocol.io/specification/2025-06-18/server/tools","note":"MCP"},{"title":"OpenAPI Specification v3.1.1","url":"https://spec.openapis.org/oas/v3.1.1.html","note":"OpenAPI Initiative"},{"title":"Web of Things Thing Description 1.1","url":"https://www.w3.org/TR/wot-thing-description11/","note":"W3C"},{"title":"JSON Schema Core","url":"https://json-schema.org/draft/2020-12/json-schema-core","note":"JSON Schema"},{"title":"JSON Schema Validation","url":"https://json-schema.org/draft/2020-12/json-schema-validation","note":"JSON Schema"},{"title":"RFC 9110: HTTP Semantics","url":"https://www.rfc-editor.org/rfc/rfc9110.html","note":"IETF"},{"title":"PROV-O: The PROV Ontology","url":"https://www.w3.org/TR/prov-o/","note":"W3C"},{"title":"MCP Authorization","url":"https://modelcontextprotocol.io/specification/2025-06-18/basic/authorization","note":"MCP"}],"openQuestions":["Restore independent external review and verify source versions, errata, licensing and claim applicability before canonical promotion.","Develop adoption schemas and tested bindings covering no-input tools, absent output schemas, duplicate names, unsupported vocabularies, partial results, stale discovery and unknown completion.","Qualify authority, effect, disclosure and retention profiles with accountable domain reviewers; test local creation, revision and retirement semantics without transferring external master ownership.","No independent external provider review; local self-audit is not independent assurance.","Direct HTTP checks are unattempted under the owner-reported sandbox block; no statuses or response hashes were measured.","No exhaustive review of later protocol versions, errata, licenses or implementation-specific extensions.","Candidate answer groups are not executable nested instance schemas; no runtime, conformance suite or tool invocation was tested.","No universal compatibility mapping across transports, asynchronous task protocols or binary/streaming payloads.","Sector-specific safety, privacy, approval and residency policies require qualified adoption review."],"resources":{"spec":"/models/wm-ai-017-tool-description/spec.yaml","agents":"/models/wm-ai-017-tool-description/AGENTS.md","source":"https://github.com/ver-cy/world-models/tree/feat/mega-model-registry/research/runs/wm-ai-017"},"provenance":{"origin":"world-models research","builtFrom":["models/wm-ai-017-tool-description/spec.yaml"],"providers":["Codex"],"researchStatus":"reviewable-draft","generatedAt":"2026-10-21T15:45:36Z","builder":"tools/build_cards.py@1.0.0"},"completeness":{"sections":{"classifiers":"filled","whatItIs":"filled","purpose":"filled","distinguishingFeatures":"derived","structure":"filled","agentConduct":"derived","ethics":"derived","owners":"filled","relations":"filled","interaction.identity":"filled","interaction.properties":"not-applicable","interaction.recognition":"missing","interaction.capabilities":"filled","interaction.hazards":"missing","interaction.interfaces":"derived","interaction.context":"filled","sources":"filled"},"notes":{"distinguishingFeatures":"Derived from boundary notes against neighbouring models.","agentConduct":"Derived from functions, policies, CRUD and access rules; prohibitions were not authored for agents as such.","ethics":"Sentences mentioning harm, privacy, consent or similar, collected from the specification.","interaction.properties":"Institutional or informational subject: no invented physical properties."},"score":0.75}}