feat(workbench): decisionsserverns lokala modell som interim-AI #441

Merged
admin merged 1 commits from feat/decision-server-ai into main 2026-11-02 16:13:09 +00:00
Owner

Medan den externa AI-API:n är pausad används decisionsserverns egna modell (Ollama på server-decision-01, hermes-qwen3.5:9b), nådd över tailnet. Ingen leverantörsnyckel, ingen per-token-kostnad, inga kunddokument lämnar flottan.

  • Ny DecisionOllamaGatewayClient med samma lokala admission (scope, allowlist, dataklass, extern-överföring, token-gränser). HTTP accepteras endast för loopback/RFC1918/tailnet.
  • createInf0Gateway prioriterar decisionsmodellen, sedan LiteLLM, sedan native INF0.
  • Evidens beräknas deterministiskt ur värdet när en liten modell utelämnar offsets: värdet måste förekomma exakt en gång i källan, annars needs_review.

Verifierat: 6 nya tester, full typecheck grön; containern på server6 når 100.96.133.104:11434.

Medan den externa AI-API:n är pausad används **decisionsserverns egna modell** (Ollama på server-decision-01, `hermes-qwen3.5:9b`), nådd över tailnet. Ingen leverantörsnyckel, ingen per-token-kostnad, inga kunddokument lämnar flottan. - Ny `DecisionOllamaGatewayClient` med samma lokala admission (scope, allowlist, dataklass, extern-överföring, token-gränser). HTTP accepteras endast för loopback/RFC1918/tailnet. - `createInf0Gateway` prioriterar decisionsmodellen, sedan LiteLLM, sedan native INF0. - Evidens beräknas deterministiskt ur värdet när en liten modell utelämnar offsets: värdet måste förekomma exakt en gång i källan, annars `needs_review`. Verifierat: 6 nya tester, full typecheck grön; containern på server6 når `100.96.133.104:11434`.
admin added 1 commit 2026-11-02 16:09:45 +00:00
feat(workbench): use the decision server's local model while the AI API is on hold
CI (SIAX Cloud) / sonar (pull_request) Skipped
CI (SIAX Cloud) / integration (push) Skipped
CI (SIAX Cloud) / sonar (push) Skipped
masterplan-lock / check (pull_request) Successful in 3s
masterplan-lock / check (push) Successful in 5s
Security Scan (SIAX Cloud) / security (push) Successful in 15s
CI (SIAX Cloud) / security (pull_request) Successful in 22s
CI (SIAX Cloud) / security (push) Successful in 23s
Security Scan (SIAX Cloud) / security (pull_request) Successful in 23s
CI (SIAX Cloud) / integration (pull_request) Successful in 30s
CI (SIAX Cloud) / quality (push) Successful in 45s
CI (SIAX Cloud) / quality (pull_request) Successful in 1m18s
525bf6b9a0
- DecisionOllamaGatewayClient calls the self-hosted Ollama on
  server-decision-01 (tailnet HTTP only for loopback/RFC1918/100.64-127),
  keeps the same local admission and reports providerKind=local.
- createInf0Gateway prefers it over INF0/LiteLLM when configured.
- Evidence is computed deterministically from values when a small model omits
  offsets; a value must occur exactly once in the source or the result goes to
  review. This makes a 9B local model usable without weakening verification.

Tests: 4 new gateway cases + 2 evidence-derivation cases; full typecheck green.
CODEOWNERS rules requested review from sax3l 2026-11-02 16:09:45 +00:00
admin merged commit f9f415f0e0 into main 2026-11-02 16:13:09 +00:00
Sign in to join this conversation.
No Reviewers
No labels
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: admin/0pt#441