Skip to content

eidolons v4.5.0

Choose a tag to compare

View all tags
@github-actions github-actions released this 05 Oct 17:44
· 6 commits to main since this release
v4.5.0
de59fa9
This commit was created on GitHub.com and signed with GitHub’s verified signature.
GPG key ID: B5690EEEBB952194
Verified
Learn about vigilant mode.

Minor: refresh OpenAI models, add explicit Claude Pro/Max profiles, and support member-specific reasoning effort defaults without changing routing tiers.

Changed

  • Refresh the OpenAI Codex profile: light → gpt-6-luna, standard → gpt-6.1-sol, deep → gpt-6-astra. Shared tier efforts remain low / medium / medium, and member tiers are unchanged. RAMZA alone defaults to high effort in the OpenAI profile for planning. Run eidolons sync in consuming projects to refresh managed descriptors and lock provenance; explicit model pins and calibration overrides retain precedence.

Added

  • Add opt-in Claude plan profiles: anthropic-pro uses Haiku / Sonnet / Opus; anthropic-max uses Haiku / Sonnet / Fable. The existing anthropic default and automatic selection stay unchanged. Document host-scoped selection, manual Max-to-Pro downgrade, and native availability fallback limits; no plan detection or fallback settings are added.
  • Profiles can set members.<id>.reasoning_effort defaults. Consumer host-member, member, and host-wide effort overrides retain precedence over profile member and tier defaults.
Assets 3
Loading