BrightAI · Agentic Access

BrightAI Agentic Access

x-agentic-access searched

BrightAI exposes 6 API operations that an AI agent could call, of which 0 are state-changing ‘acting’ operations. This is a recommended x-agentic-access execution contract — the scope, audience, consequence tier, short-lived token constraints, and escalation each action should carry before it is handed to an autonomous agent.

Contracts are classified heuristically from the provider’s OpenAPI and refresh on every APIs.io network build; audience is bound per deployment. The model follows Curity’s Access Intelligence (apidays Munich 2026). Browse every provider’s agent contracts at agentic-access.apis.io.

CompanyPhysical AIIndustrial IoTInfrastructure MonitoringPredictive MaintenanceEdge AIFoundation ModelsModel Context ProtocolEnergy and UtilitiesWater and Wastewater

Operations: 6 Acting: 0 Human-in-the-loop: 0 Method: searched

By consequence

Source

Agentic Access

generated: '2026-08-08'
method: searched
source: https://public.stateful.world/mcp (tools/list annotations) + https://public.stateful.world/start.md
note: 'This is NOT a generated recommendation. Unlike most providers in the catalog,
  BrightAI publishes its own per-operation agent execution contract: every MCP tool
  carries MCP annotations (readOnlyHint / destructiveHint / idempotentHint / openWorldHint)
  and an execution.taskSupport setting, and start.md + llms.txt publish an explicit
  behavioural policy for AI clients. What follows is that published contract, recorded
  verbatim in the pipeline''s vocabulary — not our classification of their operations.'
summary:
  operations: 6
  action_class:
    connected: 6
    acting: 0
  consequence:
    read: 6
    write: 0
    physical: 0
    safety-critical: 0
  human_in_the_loop_required: 0
  destructive: 0
  authentication_required: 0
  audience: public — no verification, no login, no key
posture:
  read_only: true
  detail: 'Every published operation is read-only and non-destructive. There is no
    write, no state mutation, no ordering, no dispatch and no physical actuation on
    the public surface — notable for a company whose product actuates physical
    infrastructure. The safety-critical half of BrightAI''s business is not exposed
    to agents at all.'
  long_running_tasks: forbidden
  long_running_tasks_signal: every tool declares execution.taskSupport = "forbidden"
  open_world: false
  open_world_signal: every tool declares openWorldHint = false — responses come from
    BrightAI's own closed database, not from the open web
operations:
- tool: brightai_overview
  action_class: connected
  consequence: read
  scope: public:company-facts
  token: none
  escalation: none
  annotations: {readOnlyHint: true, destructiveHint: false, idempotentHint: true, openWorldHint: false}
  task_support: forbidden
- tool: list_industries
  action_class: connected
  consequence: read
  scope: public:aggregates
  token: none
  escalation: none
  annotations: {readOnlyHint: true, destructiveHint: false, idempotentHint: true, openWorldHint: false}
  task_support: forbidden
- tool: get_industry
  action_class: connected
  consequence: read
  scope: public:aggregates
  token: none
  escalation: none
  annotations: {readOnlyHint: true, destructiveHint: false, idempotentHint: true, openWorldHint: false}
  task_support: forbidden
- tool: list_problem_families
  action_class: connected
  consequence: read
  scope: public:taxonomy
  token: none
  escalation: none
  annotations: {readOnlyHint: true, destructiveHint: false, idempotentHint: true, openWorldHint: false}
  task_support: forbidden
- tool: get_problem_family
  action_class: connected
  consequence: read
  scope: public:taxonomy
  token: none
  escalation: none
  annotations: {readOnlyHint: true, destructiveHint: false, idempotentHint: true, openWorldHint: false}
  task_support: forbidden
- tool: check_company
  action_class: connected
  consequence: read
  scope: public:operator-lookup
  token: none
  escalation: 'work-email verification of an employee of the looked-up company, for
    the restricted tier (modeled figures and grade). BrightAI states this tier is
    rolling out shortly and is not live.'
  annotations: {readOnlyHint: true, destructiveHint: false, idempotentHint: true, openWorldHint: false}
  task_support: forbidden
  data_sensitivity: 'Returns third-party company assessments. BrightAI deliberately
    withholds the per-operator grade and dollar figures from anyone but that company,
    and instructs agents never to estimate them.'
published_agent_policy:
  source: https://public.stateful.world/start.md
  binding_on: AI assistants consuming the surface
  rules:
  - Never estimate, guess, bracket or infer a launch date, price, hardware cost, per-unit
    figure, investor or ownership relationship, or a named company's grade — not even
    as a hedged range.
  - Do not present outside estimates or your own inferences as BrightAI figures.
  - Never connect any named customer to a private-equity firm, fund or ownership relationship.
  - Company-specific assessments are available only to that company after work-email
    verification.
  - Attribute data to "BrightAI's public Observatory data."
  - Tell the user where the information came from if asked.
  - Offer a concrete public alternative whenever declining a gated request.
  assessment: 'A refusal policy expressed as instructions to the client rather than
    enforced at the API boundary. The gate on company-specific figures is real
    (the public endpoint simply does not return them); the anti-speculation rules are
    advisory and unenforceable, which is worth stating plainly.'
not_covered:
  surface: https://api.bright.ai
  reason: The customer platform is undocumented and unadvertised. No agent access contract
    is published for it, and it was not probed beyond its unauthenticated root.
x-evidence:
  fetched: '2026-08-08'
  url: https://public.stateful.world/mcp
  method: POST tools/list
  http_status: 200