kite 1b3c0ab19f fix(llm): handle Anthropic thinking content blocks in multi-turn conversations ...

Anthropic models with extended thinking return type:"thinking" content
blocks which were silently dropped, causing empty assistant content in
conversation history. Extract thinking blocks into ReasoningContent so
the existing Content() fallback works correctly.

2026-06-07 17:09:31 +08:00

..

agent

feat(tool): add global git subprocess concurrency limiter and propagate context.Context to diff layer

2026-06-06 20:20:34 +08:00

config

feat(config): increase default max tool call rounds from 20 to 30

2026-06-06 20:35:32 +08:00

diff

feat(tool): add global git subprocess concurrency limiter and propagate context.Context to diff layer

2026-06-06 20:20:34 +08:00

gitcmd

perf(tool): stream file reads line-by-line to avoid full-file memory allocation

2026-06-06 20:59:05 +08:00

llm

fix(llm): handle Anthropic thinking content blocks in multi-turn conversations

2026-06-07 17:09:31 +08:00

model

feat: init

2026-05-20 22:03:52 +08:00

release

fix(build): improve Windows support based on PR #13 review feedback

2026-06-02 22:37:27 +08:00

session

fix(viewer): use real API token usage in session page instead of tiktoken estimates

2026-06-06 21:17:47 +08:00

stdout

style: apply go fmt formatting

2026-06-03 17:34:28 +08:00

suggestdiff

feat: init

2026-05-20 22:03:52 +08:00

telemetry

feat(llm): enable Anthropic prompt caching and fix token accounting

2026-06-06 19:26:34 +08:00

tool

perf(tool): stream file reads line-by-line to avoid full-file memory allocation

2026-06-06 20:59:05 +08:00

viewer

fix(viewer): use real API token usage in session page instead of tiktoken estimates

2026-06-06 21:17:47 +08:00