AIJIM PROTOOLS
LINACurrent

LINA capabilities

Current assistant boundaries after the Core-authority and honest-absence hardening.

Capability boundary

LINA is an interaction and orchestration layer. Core remains the write authority for audit, signer, publication, review, and durable story state. When a capability is not runtime-backed, the UI must say so instead of returning fabricated success.

Runtime-backed today

Stage navigation

LINA may help users move between Setup, Work, Create, Team, story reader, evidence graph, and settings routes without inventing a separate visual model.

Claim and evidence explanation

LINA can explain current story state, claim review progress, evidence bindings, and verifier-readiness when those values are present in the workspace context.

Bounded retrieval

RAG search is routed through the Core-owned service path. Direct authenticated RPC access is closed; the app proxy carries the tenant boundary.

Action discovery

LINA may suggest available next actions, but deferred tools remain deferred and cannot be recommended as if they were runtime-backed.

Voice identity

Voice output is an interaction layer. It does not own audit state, publication authority, or claim truth.

Explicitly not claimed

  • 3D universe navigation as a shipped 1.0 surface
  • Direct story, signer, review-gate, or impact writes from LINA
  • Hash-verified impact metrics before server-attested aggregation exists
  • Publication approval without explicit human and Core review gates
  • Any deferred placeholder tool returning a successful runtime result

Confirmation gates

LINA can surface these actions, but cannot silently execute them. The product keeps the human/Core boundary visible at every durable write point.

Publishing or regenerating a publication bundle
Writing signer-registry entries
Running batch analysis that becomes durable evidence
Changing workspace membership, roles, or API-key state
Any action that would cross from read/explain into durable write authority