slug: arthurai provider: ArthurAI generated_by: planning/capability-mapping/scripts/classify_capabilities.py model: claude-opus-5 frame: - Software & Technology min_confidence: 0.7 capability_model: source: https://github.com/vincentmakes/turbo-ea-capabilities license: CC-BY-4.0 attribution: Turbo EA Capabilities by Vincent Verdet — Turbo EA, https://github.com/vincentmakes/turbo-ea-capabilities, CC BY 4.0 notice: NOTICE edge_count: 19 edges: - tag: Authorization V1 spec_file: arthurai-authorization-v1-api-openapi.yml capability_id: BC-620.20 capability_id_l1: BC-620 capability_name: Identity & Access Management confidence: 0.85 evidence: POST /api/v1/organization/role_bindings Post Organization Role Binding; GET /api/v1/permissions List Permissions; POST /api/v1/permissions/check Check Permissions reason: Role bindings, roles, permission checks across org/workspace/project — RBAC authorization, i.e. identity and access management. - tag: Models V1 spec_file: arthurai-models-v1-api-openapi.yml capability_id: BC-610.60 capability_id_l1: BC-610 capability_name: Artificial Intelligence Management confidence: 0.85 evidence: '"Create Model", "Update Model Metric Configuration By Model Id", "Update Model Metrics Schedule", schema ModelProblemType' reason: CRUD and monitoring configuration over AI/ML model registrations — squarely AI/ML model lifecycle management. - tag: LLMEvals spec_file: arthurai-llmevals-api-openapi.yml capability_id: BC-610.60 capability_id_l1: BC-610 capability_name: Artificial Intelligence Management confidence: 0.82 evidence: '"Save an llm eval", "Run a saved llm eval", "List all versions of an llm eval" — LLM-as-a-judge evaluation lifecycle for AI models' reason: Operations create, version and execute LLM evaluations, which is model evaluation/validation within the AI/ML model lifecycle (MLOps, responsible AI). - tag: Users V1 spec_file: arthurai-users-v1-api-openapi.yml capability_id: BC-620.20 capability_id_l1: BC-620 capability_name: Identity & Access Management confidence: 0.8 evidence: POST /api/v1/organization/users/invite-user Create User Via Email Invite; Create Service Account; Regenerate User Credentials; PostGlobalRoleBinding reason: Operations create, invite, patch and delete organization users and service accounts, manage credentials and role bindings — classic identity and access management. - tag: Agents V1 spec_file: arthurai-agents-v1-api-openapi.yml capability_id: BC-610.60 capability_id_l1: BC-610 capability_name: Artificial Intelligence Management confidence: 0.78 evidence: GET /api/v1/organization/unregistered_agents "List Organization Unregistered Agents"; GET /api/v1/workspaces/{workspace_id}/agents/registered/llm_models; schemas AgentResponse, DataSourceResponse, OTELAgentCreationSource reason: Registry of AI agents with their LLM models, tools and data sources, plus registration state — AI inventory and governance over the model/agent lifecycle. - tag: Default Validation spec_file: arthurai-default-validation-api-openapi.yml capability_id: BC-610.60 capability_id_l1: BC-610 capability_name: Artificial Intelligence Management confidence: 0.78 evidence: '''Endpoints to validate prompt and response on default rules''; schemas PIIDetailsResponse, ToxicityDetailsResponse, HallucinationDetailsResponse, RuleResultEnum' reason: Runtime guardrail validation of LLM prompts and responses for PII, toxicity and hallucination is responsible-AI control within AI/ML model lifecycle management. - tag: ML Evals spec_file: arthurai-ml-evals-api-openapi.yml capability_id: BC-610.60 capability_id_l1: BC-610 capability_name: Artificial Intelligence Management confidence: 0.78 evidence: '"Save an ML eval" with schemas PIIEvalConfig, ToxicityEvalConfig, PromptInjectionEvalConfig' reason: Single operation registers ML evaluation configurations (toxicity, PII, prompt injection) — responsible-AI model evaluation, part of AI/ML lifecycle management. - tag: Metrics V1 spec_file: arthurai-metrics-v1-api-openapi.yml capability_id: BC-610.60 capability_id_l1: BC-610 capability_name: Artificial Intelligence Management confidence: 0.78 evidence: '"Query Model Metrics", "Upload Metrics For Version", "Create A Metric Version" under /api/v1/models/{model_id}/metrics' reason: Model-scoped metric versioning, ingestion and querying is ML model performance monitoring — AI/ML model lifecycle (MLOps) rather than generic IT monitoring. - tag: Groups V1 spec_file: arthurai-groups-v1-api-openapi.yml capability_id: BC-620.20 capability_id_l1: BC-620 capability_name: Identity & Access Management confidence: 0.76 evidence: GET /api/v1/organization/groups 'List Groups'; POST /api/v1/groups/{group_id}/users 'Assign User To Group'; DELETE 'Remove Users From Group In Bulk'; schemas GroupMembership, UserType reason: Group and group-membership administration for organization users is access-control/identity administration. Could alternatively be read as tenant user administration, hence not higher. - tag: Continuous Evals spec_file: arthurai-continuous-evals-api-openapi.yml capability_id: BC-610.60 capability_id_l1: BC-610 capability_name: Artificial Intelligence Management confidence: 0.75 evidence: POST /api/v1/tasks/{task_id}/continuous_evals Create a continuous eval; GET .../continuous_evals/results Get all continuous eval run results reason: Definition and running of continuous LLM/agent evaluations with results and analytics — AI model lifecycle evaluation/monitoring (responsible AI). - tag: Spans spec_file: arthurai-spans-api-openapi.yml capability_id: BC-4220.20 capability_id_l1: BC-4220 capability_name: Observability Management confidence: 0.75 evidence: '"Query Traces", "Compute Metrics for Span", "List Span Metadata with Filtering"; schemas SpanWithMetricsResponse, QueryTracesWithMetricsResponse' reason: Spans and traces with computed metrics are the core of distributed-tracing observability; the operations query and enrich telemetry from the running service, matching Observability Management. - tag: Task Based Validation spec_file: arthurai-task-based-validation-api-openapi.yml capability_id: BC-610.60 capability_id_l1: BC-610 capability_name: Artificial Intelligence Management confidence: 0.75 evidence: '"Endpoints to validate prompt and response for a task"; Validate Prompt Endpoint / Validate Response Endpoint' reason: Runtime guardrail validation of GenAI prompts and responses against configured rules is responsible-AI governance, mapped to Artificial Intelligence Management. - tag: Prompts spec_file: arthurai-prompts-api-openapi.yml capability_id: BC-610.60 capability_id_l1: BC-610 capability_name: Artificial Intelligence Management confidence: 0.72 evidence: '"Save an agentic prompt", "List all versions of an agentic prompt", "Run/Stream a specific version of an agentic prompt"; schemas LLMConfigSettings, ModelProvider' reason: Operations manage the lifecycle (save, version, tag, run, render) of LLM prompts with model provider configuration — this is AI/LLM asset lifecycle management, best matched by Artificial Intelligence Management. Not a generic content or product-spec capability. - tag: Rules spec_file: arthurai-rules-api-openapi.yml capability_id: BC-610.60 capability_id_l1: BC-610 capability_name: Artificial Intelligence Management confidence: 0.72 evidence: '"Endpoints to manage rules", "Create Default Rule"; schemas ToxicityConfig, PIIConfig, RuleScope, RuleType' reason: These are guardrail rules (PII, toxicity, regex, keywords) that validate LLM prompts and responses — responsible-AI controls within AI model lifecycle governance, not enterprise policy or compliance management. - tag: Stateless Validation spec_file: arthurai-stateless-validation-api-openapi.yml capability_id: BC-610.60 capability_id_l1: BC-610 capability_name: Artificial Intelligence Management confidence: 0.72 evidence: POST /api/v2/validate Stateless Validate; schemas PIIConfig, ToxicityConfig, HallucinationDetailsResponse, RuleResultEnum reason: Endpoint applies guardrail rules (PII, toxicity, hallucination) to LLM prompts/responses — responsible-AI control enforcement, which falls under AI/ML lifecycle and responsible AI management rather than any generic capability. - tag: Agent Discovery spec_file: arthurai-agent-discovery-api-openapi.yml capability_id: BC-610.60 capability_id_l1: BC-610 capability_name: Artificial Intelligence Management confidence: 0.7 evidence: 'POST /api/v1/agent-polling/execute-all "Execute All Agent Polling"; schema DiscoverAndPollResponse; vendor: "agent discovery across OTEL/MCP/network/cloud APIs"' reason: Discovery/polling of AI agents deployed in the estate is part of AI/ML inventory and governance (responsible AI oversight), the closest supported capability. - tag: Alerts V1 spec_file: arthurai-alerts-v1-api-openapi.yml capability_id: BC-4220.20 capability_id_l1: BC-4220 capability_name: Observability Management confidence: 0.7 evidence: 'GET /api/v1/models/{model_id}/alerts Get Model Alerts; schemas: Alert, AlertRuleInterval, AlertWebhookCalled, AlertLogStatus' reason: Alerting on monitored AI models on an observability platform — monitoring/observability of running services, not financial-crime alerting. Could arguably be AI model monitoring (BC-610.60) but the operations are generic alert rules/logs, so observability is the closest defensible fit. - tag: Continuous Eval Test Runs spec_file: arthurai-continuous-eval-test-runs-api-openapi.yml capability_id: BC-610.60 capability_id_l1: BC-610 capability_name: Artificial Intelligence Management confidence: 0.7 evidence: 'POST /api/v1/continuous_evals/{eval_id}/test_runs Create and start a test run for a continuous eval; schemas: TestRunStatus, AgenticAnnotationResponse, EvalType' reason: Executing evaluation test runs against LLM/agent outputs is part of AI/ML model lifecycle evaluation and governance; not general software test management since the subject is AI models. - tag: Prompt Experiments spec_file: arthurai-prompt-experiments-api-openapi.yml capability_id: BC-610.60 capability_id_l1: BC-610 capability_name: Artificial Intelligence Management confidence: 0.7 evidence: '"Create and run a prompt experiment", "Get experiment test cases", "Get prompt results"; schema PromptEvalResultSummaries' reason: Designing and running prompt experiments with test cases and eval results is AI model/prompt experimentation within the AI-ML lifecycle, not product feature A/B testing.