Skip to content
This repository was archived by the owner on Jun 1, 2026. It is now read-only.

v0.1.5

Choose a tag to compare

View all tags
@github-actions github-actions released this 24 Apr 06:59
· 538 commits to main since this release
v0.1.5
a0ed6d7
This commit was created on GitHub.com and signed with GitHub’s verified signature.
GPG key ID: B5690EEEBB952194
Verified
Learn about vigilant mode.

Changelog

  • 9acb709 Address latest review findings
  • b8641fc Address review feedback on perf coverage
  • b50dbc9 Ensure codex completed events always publish usage stats
  • 5197cf0 Ensure codex streaming requests force stream mode
  • cf47550 Fix codex completed-event reader and stress test gating
  • 6487db8 Fix codex request cache and SSE drain regressions
  • 86e4e1e Fixed: router-for-me#2274
  • db66efc Force stream mode in codex streaming fallback path
  • 81ce655 Handle Codex capacity errors as retryable
  • f2cb22e Ignore non-auth JSON files and rename usage snapshot
  • 31026cd Optimize auth hot paths and add TTFT stress coverage
  • cfc0e3f Optimize codex request preparation under extreme load
  • dbfde0c Preserve default transport settings for proxy clients
  • a0ed6d7 backport: auth, codex, claude, antigravity, and routing fixes (#28)
  • 0f28349 backport: upstream healthz, websocket client IP, and api-call proxy fixes (#27)
  • 2420ffe build: add freebsd support to releases
  • f487dea docs(readme): update QQ group information in Chinese README
  • 823332c feat(api): support batch auth file upload and delete
  • 4ba0f75 feat(codex): pass through codex client identity headers
  • aeae747 feat(openai-compat): add per-model thinking support
  • bf06e43 fix(auth): prevent stale ModelStates inheritance from disabled auth entries
  • 398b0f5 fix(auth): restore filestore project id hydration
  • 2ed3d8a fix(codex): harden websocket transcript repair
  • 5c23abf fix(codex): strip stream_options from Responses API requests
  • eaf0ba6 fix(openai): make websocket tool repair concurrency-safe
  • f5f70a6 fix(openai-compat): preserve legacy thinking passthrough
  • 9c15d17 fix(review): preserve bootstrap headers, honor NO_PROXY, and harden integration test
  • 233411b fix(review): unify usage reporter finalization across executors
  • 3904e31 fix(security): harden management panel asset updater
  • ce53a53 fix(security): require API keys for Gemini CLI routes (#26)
  • 8210c76 fix(security): validate auth file names to prevent unsafe input
  • 5833fb3 fix(service): sync external auth updates into model registry
  • 2d897a7 fix(usage): persist stats across config changes
  • 1aa3926 fix(usage): preserve client IP in snapshot dedup
  • ffddc82 fix: backport upstream responses, gemini, auth, and codex websocket fixes onto origin/main
  • e60452f fix: finalize upstream v6.9.3 backport integration
  • 47d1d12 fix: keep a fallback turn for system-only Claude inputs
  • b4efc3c fix: map OpenAI system messages to Claude top-level system
  • bb99f7b fix: normalize model name in TranslateRequest fallback to prevent prefix leak
  • 02a486c fix: reject oversized downloads instead of truncating; warn on unverified fallback
  • c735c59 fix: 统一 usage 统计与 HTTP 日志的客户端 IP 口径
  • 6705f69 refactor(config): replace auto-update-panel with disable-auto-update-panel for clarity
  • ddf062c 持久化使用统计并支持自动恢复
Assets 11
Loading