What's Changed

• Fix: capture redacted_thinking blocks in Anthropic streaming by @vietanh85 in #94
• fix: preserve split tool call stream metadata by @slegarraga in #96
• test(openaicompat): cover split tool-call stream patch lines by @vietanh85 in #97
• feat: expose retry steps via RetryObserver (#95) by @vietanh85 in #98
• chore: switch zenflow-review to FPT provider by @vietanh85 in #99

New Contributors

• @slegarraga made their first contribution in #96

Full Changelog: v0.8.3...v0.8.4

What's Changed

• fix: detect duplicate tool names in buildToolMap by @bclermont in #91
• provider/ollama: replace compat layer with native Ollama API + think mode support by @bclermont in #92

New Contributors

• @bclermont made their first contribution in #91

Full Changelog: v0.8.2...v0.8.3

What's Changed

• fix(provider): populate ToolName on synthetic orphan tool results for… by @nehmeroumani in #90

Full Changelog: v0.8.1...v0.8.2

What's Changed

• fix(openai): handle nested error fields in Responses API stream error… by @nehmeroumani in #88
• fix(openai): preserve flat error fields when nested error is partial by @vietanh85 in #89

New Contributors

• @nehmeroumani made their first contribution in #88

Full Changelog: v0.8.0...v0.8.1

What's Changed

• feat(mcp): native OAuth 2.1 + PKCE for remote MCP servers by @DevLumuz in #80
• fix: remove committed binaries and block them in pre-commit by @vietanh85 in #84
• ci: switch zenflow-review to Google Gemini (gemini-3-flash-preview) by @vietanh85 in #86
• feat: surface hook and predicate panics instead of swallowing them by @vietanh85 in #85
• feat: add NewTool typed constructor for tool definitions by @vietanh85 in #87

New Contributors

• @DevLumuz made their first contribution in #80

Full Changelog: v0.7.11...v0.8.0

What's Changed

• feat: add llama.cpp server provider by @vietanh85 in #77

Full Changelog: v0.7.10...v0.7.11

What's Changed

• feat: add llama.cpp server provider by @vietanh85 in #77

Full Changelog: v0.7.10...v0.7.11

What's Changed

• feat: add llama.cpp server provider by @vietanh85 in #77

Full Changelog: v0.7.10...v0.7.11

What's Changed

• feat: add llama.cpp server provider by @vietanh85 in #77

Full Changelog: v0.7.10...v0.7.11

What's Changed

• fix(mcp): HTTPTransport tolerates POST-only Streamable HTTP servers (#74) by @vietanh85 in #76

Full Changelog: v0.7.9...v0.7.10

What's Changed

• fix(openai): handle Responses API SSE lines larger than 1MiB (#70) by @vietanh85 in #75

Full Changelog: v0.7.8...v0.7.9

What's Changed

• fix(openaicompat): echo reasoning_content on assistant messages by @tenjinfactory in #72
• fix(sse): handle SSE lines larger than 1MiB (#70) by @vietanh85 in #73

New Contributors

• @tenjinfactory made their first contribution in #72

Full Changelog: v0.7.7...v0.7.8

What's Changed

• ci(zenflow-review): add comment-triggered AI PR review workflow by @vietanh85 in #64
• fix(zenflow-review): pass --sandbox to grant tool permissions by @vietanh85 in #66
• docs(contributing): mention /zenflow-review trigger by @vietanh85 in #65
• fix(openai): rename max_tokens by model, not reasoning_effort by @vietanh85 in #69

Full Changelog: v0.7.6...v0.7.7