University of Manchester · Agentic Access

University of Manchester Agentic Access

x-agentic-access derived

University of Manchester exposes 4 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.

By consequence: 4 read.

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.

UniversityHigher EducationEducationResearchUnited KingdomRussell GroupLibraryDigital CollectionsIIIFIdentity FederationResearch DataResearch Computing

Operations: 4 Acting: 0 Human-in-the-loop: 0 Method: derived

By consequence

read 4 write 0 safety-critical 0

Source

Agentic Access

---
generated: '2026-08-19'
method: derived
source: >-
  Derived from the two institution-operated contracts in openapi/ (Manchester Digital Collections
  IIIF Presentation and Image APIs). Replaces a previous artifact that classified 951 operations
  from Elsevier's Pure specification and is now quarantined.
x-operator: institution
description: >-
  Recommended x-agentic-access execution contracts for the surfaces the University of Manchester
  actually operates. Both are anonymous, read-only, cacheable public-heritage APIs — about as safe a
  target for autonomous agents as exists — so the governance question here is not consequence but
  courtesy: there is no published rate-limit table and no status page, and an agent that walks 398
  canvases per manuscript across 61 manuscripts in one collection can trivially outrun a library
  image server.
summary:
  operations: 4
  by_action_class:
    connected: 4
    acting: 0
  by_consequence:
    read: 4
    write: 0
    safety-critical: 0
  human_in_the_loop_required: 0
  authentication_required: false
operations:
- path: /iiif/{documentId}
  method: get
  operationId: getManifest
  surface: iiif-presentation-api
  x-agentic-access:
    action-class: connected
    consequence: read
    audience: public
    human-in-the-loop: false
    idempotent: true
    cacheable: true
    notes: >-
      Manifests are large — the verified example was 378 KB for 398 canvases. Agents should cache by
      documentId and avoid re-fetching a manifest to reach a single canvas.
- path: /iiif/collection/{collectionId}
  method: get
  operationId: getCollection
  surface: iiif-presentation-api
  x-agentic-access:
    action-class: connected
    consequence: read
    audience: public
    human-in-the-loop: false
    idempotent: true
    cacheable: true
    notes: The correct entry point for discovery; enumerate here rather than guessing document ids.
- path: /iiif/{identifier}/info.json
  method: get
  operationId: getImageInfo
  surface: iiif-image-api
  x-agentic-access:
    action-class: connected
    consequence: read
    audience: public
    human-in-the-loop: false
    idempotent: true
    cacheable: true
- path: /iiif/{identifier}/{region}/{size}/{rotation}/{quality}.{format}
  method: get
  operationId: getImage
  surface: iiif-image-api
  x-agentic-access:
    action-class: connected
    consequence: read
    audience: public
    human-in-the-loop: false
    idempotent: true
    cacheable: true
    notes: >-
      Bounded by the server's own maxWidth/maxHeight of 2000. Prefer the pre-generated sizes listed
      in info.json over arbitrary derivative requests; each novel region/size combination is work the
      server has to do.
guidance:
  rate_limits: >-
    None published and none observed in response headers. Treat as a shared academic resource:
    serialize requests, honour any Retry-After, and contact the Library before bulk harvesting.
  reuse_terms: https://www.digitalcollections.manchester.ac.uk/terms/
  attribution_required: >-
    Yes. Every manifest carries an attribution string naming the University of Manchester and the
    University of Manchester Library, and agents reproducing content must carry it through.
excluded:
- surface: Elsevier Pure CRIS
  reason: >-
    Tenant-operated. The contract, the api-key scheme and the operation set are Elsevier's; agentic
    governance for it belongs against Elsevier's own profile, not Manchester's.
- surface: Shibboleth SAML Identity Provider
  reason: >-
    Not an agent-callable API. An assertion cannot be obtained without service-provider registration
    in the UK Access Management Federation.