Problem: AI-provider API keys are read directly from the operator's own environment inside the self-host container today and never touch the central Worker at all — centralizing them for a hosted model is fully greenfield work.
Area: ORB / Security
Proposal: Design a path for AI-provider keys to be held per-installation centrally, extending the existing per-installation credential-isolation pattern already proven safe for GitHub tokens (see the extension issue below).
Deliverables:
- A written design, then an implementation, for centrally-held AI-provider keys per installation.
Cross-reference (added 2026-07-12)
See also #5120 ("Optional Infisical integration for self-host secrets management") — a separate, already-filed self-host-side secrets plan. Read it before finalizing centralized custody design here, so self-host and hosted don't end up with two divergent secrets stories.
Problem: AI-provider API keys are read directly from the operator's own environment inside the self-host container today and never touch the central Worker at all — centralizing them for a hosted model is fully greenfield work.
Area: ORB / Security
Proposal: Design a path for AI-provider keys to be held per-installation centrally, extending the existing per-installation credential-isolation pattern already proven safe for GitHub tokens (see the extension issue below).
Deliverables:
Cross-reference (added 2026-07-12)
See also #5120 ("Optional Infisical integration for self-host secrets management") — a separate, already-filed self-host-side secrets plan. Read it before finalizing centralized custody design here, so self-host and hosted don't end up with two divergent secrets stories.