hub-gif 34e448728f fix(llm): 剥除混在 content 里的思考标签，保留正文 ...

normalize_message_content 经 strip_thinking_leaks；支持 MA_LLM_PRESERVE_THINKING_IN_OUTPUT；修复 text 凭据单测在含 .env 时的干扰。

Made-with: Cursor

2026-04-27 14:04:19 +08:00

..

__init__.py

fix(llm): 剥除混在 content 里的思考标签，保留正文

2026-04-27 14:04:19 +08:00

__main__.py

refactor: 删除 AI_crawler 薄壳，配料试跑用 python -m pipeline.openai_gateway

2026-04-27 10:26:49 +08:00

chat_content.py

fix(llm): 剥除混在 content 里的思考标签，保留正文

2026-04-27 14:04:19 +08:00

constants.py

refactor(pipeline): OpenAI 网关与配料视觉抽离至 openai_gateway

2026-04-27 10:24:50 +08:00

credentials.py

feat(openai_gateway): 纯文本可单独使用 OPENAI_TEXT_* 凭据

2026-04-27 11:33:14 +08:00

estimate.py

refactor(pipeline): OpenAI 网关与配料视觉抽离至 openai_gateway

2026-04-27 10:24:50 +08:00

ingredients_defaults.py

refactor(pipeline): OpenAI 网关与配料视觉抽离至 openai_gateway

2026-04-27 10:24:50 +08:00

ingredients_op.py

refactor(pipeline): OpenAI 网关与配料视觉抽离至 openai_gateway

2026-04-27 10:24:50 +08:00

text_chat.py

feat(openai_gateway): 纯文本可单独使用 OPENAI_TEXT_* 凭据

2026-04-27 11:33:14 +08:00

timeouts.py

refactor(pipeline): OpenAI 网关与配料视觉抽离至 openai_gateway

2026-04-27 10:24:50 +08:00