zhayujieandcowagent 896e76a455 feat: enable prompt caching for Claude agent requests ...

Cache the system prompt and conversation prefix on tool-carrying requests,
with a 1h TTL on the shared system + tools prefix (configurable via
claude_cache_ttl) and 5m on the conversation tail. Report prompt_tokens
as the full prompt including cached tokens.

Co-authored-by: cowagent <cow@cowagent.ai>

2026-10-04 11:35:30 +08:00

..

baidu

Merge pull request #3444 from c020627/fix-baidu-token-not-logged

2026-10-03 19:08:36 +08:00

chatgpt

fix(chatgpt): fall back to the openai config values for azure dalle

2026-10-03 05:28:26 +05:30

claudeapi

feat: enable prompt caching for Claude agent requests

2026-10-04 11:35:30 +08:00

dashscope

fix(dashscope): keep the system prompt a session is built with

2026-09-29 18:29:54 +08:00

deepseek

fix: share one failure-body reader across the reply_text bots

2026-10-03 18:49:04 +08:00

doubao

fix: share one failure-body reader across the reply_text bots

2026-10-03 18:49:04 +08:00

gemini

feat: add gemini-3.8-flash model

2026-09-10 12:02:48 +08:00

linkai

fix: trim batch-6 comments and tests, clean up kf voice segments

2026-10-03 23:31:54 +08:00

mimo

Merge pull request #3128 from Anai-Guo/fix/except-closure-error-generator

2026-09-09 16:04:51 +08:00

minimax

style(minimax): drop the placeholder-free f prefixes the ruff gate flags

2026-10-02 19:33:32 +05:30

modelscope

fix: trim batch-6 comments and tests, clean up kf voice segments

2026-10-03 23:31:54 +08:00

moonshot

fix: share one failure-body reader across the reply_text bots

2026-10-03 18:49:04 +08:00

openai

fix: remove unused import and redundant f-string flagged by ruff

2026-09-28 16:25:12 +08:00

qianfan

feat(models): add baidu ernie-5.1

2026-05-10 18:39:38 +08:00

xunfei

fix(voice): verify the Xunfei Spark websocket certificate

2026-10-03 05:33:06 +05:30

zhipuai

feat(models): surface real provider token usage in streaming responses

2026-09-06 18:31:15 +08:00

bot_factory.py

fix(models): resolve custom credentials from the routed provider, not global

2026-09-11 15:21:20 +00:00

bot.py

fix: share one failure-body reader across the reply_text bots

2026-10-03 18:49:04 +08:00

custom_provider.py

fix(models): resolve custom credentials from the routed provider, not global

2026-09-11 15:21:20 +00:00

model_catalog.py

fix: fold the capability case fix into one line and the catalog tests

2026-10-03 18:55:25 +08:00

openai_compatible_bot.py

fix: trim batch-6 comments and tests, clean up kf voice segments

2026-10-03 23:31:54 +08:00

reasoning_capabilities.py

fix(models): treat qwen3.8-max/flash as hybrid thinking models

2026-09-16 12:33:53 +08:00

session_manager.py

chore: the bot directory was changed to models

2026-02-01 15:21:28 +08:00