Grayscale/nanobot
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity
949 Commits 1 Branch 1 Tag
4ba8d137bc414b31776e2d79abb2d585e36a66bb
Commit Graph
56 Commits
Author SHA1 Message Date
Re-bin 3c79404194 fix(providers): sanitize thinking_blocks by provider and harden content normalization 2026-03-02 06:58:10 +00:00
Jack LuandGitHub 3ee061b879 Merge branch 'main' into main 2026-03-01 13:35:24 +08:00
Re-bin 5ca386ebf5 fix: preserve reasoning_content and thinking_blocks in session history 2026-02-28 17:37:12 +00:00
Re-bin f9d72e2e74 feat: add reasoning_effort config to enable LLM thinking mode 2026-02-28 17:18:05 +00:00
JK_Lu 977ca725f2 style: unify code formatting and import order 
- Remove trailing whitespace and normalize blank lines
- Unify string quotes and line breaks for long lines
- Sort imports alphabetically across modules
 2026-02-28 20:55:43 +08:00
Re-bin 1fe94898f6 fix: generate short alphanumeric tool_call_id for Mistral compatibility 2026-02-27 16:13:26 +00:00
haosenwang1018 8de2f8d588 fix: preserve reasoning_content in message sanitization for thinking models 
_sanitize_messages strips all non-standard keys from messages, including
reasoning_content. Thinking-enabled models like Moonshot Kimi k2.5
require reasoning_content to be present in assistant tool call messages
when thinking mode is on, causing a BadRequestError (#1014).

Add reasoning_content to _ALLOWED_MSG_KEYS so it passes through
sanitization when present.

Fixes #1014
 2026-02-24 04:21:55 +08:00
Re-bin b653183bb0 refactor(providers): move empty content sanitization to base class 2026-02-22 18:26:42 +00:00
andienguyen-ecoligo de63c31d43 fix(providers): normalize empty reasoning_content to None at provider level 
PR #947 fixed the consumer side (context.py) but the root cause is at
the provider level — getattr returns "" (empty string) instead of None
when reasoning_content is empty. This causes DeepSeek API to reject the
request with "Missing reasoning_content field" error.

`"" or None` evaluates to None, preventing empty strings from
propagating downstream.

Fixes #946
 2026-02-21 12:30:57 -05:00
Re-bin cc04bc4dd1 fix: check gateway's supports_prompt_caching instead of always returning False 2026-02-20 15:14:45 +00:00
Re-bin 5cc019bf1a style: trim verbose comments in _sanitize_messages 2026-02-20 11:27:21 +00:00
Re-bin 0c2fea6d33 Merge branch 'main' into pr-795 2026-02-20 11:25:51 +00:00
Re-bin 9ffae47c13 refactor(litellm): remove redundant comments in cache_control methods 2026-02-20 08:21:02 +00:00
Re-bin afa0513243 Merge branch 'main' into pr-854 2026-02-20 08:17:32 +00:00
tercerapersonaandClaude Sonnet 4.6 3b4763b3f9 feat: add Anthropic prompt caching via cache_control 
Inject cache_control: {"type": "ephemeral"} on the system message and
last tool definition for providers that support prompt caching. Adds
supports_prompt_caching flag to ProviderSpec (enabled for Anthropic only)
and skips caching when routing through a gateway.

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
 2026-02-19 11:05:22 -03:00
PiEgg 9789307dd6 Fix Codex provider routing for GitHub Copilot models 2026-02-19 13:30:02 +08:00
Ivan e44f14379a fix: sanitize messages and ensure 'content' for strict LLM providers 
- Strip non-standard keys like 'reasoning_content' before sending to LLM
- Always include 'content' key in assistant messages (required by StepFun)
- Add _sanitize_messages to LiteLLMProvider to prevent 400 BadRequest errors
 2026-02-18 11:57:58 +03:00
Re-bin 9e5f7348fe Merge branch 'main' into pr-151 2026-02-16 09:19:40 +00:00
Re-bin 49fec3684a fix: use json_repair for robust LLM response parsing 2026-02-15 08:11:33 +00:00
The Mavik 10e9e0cdc9 fix(providers): clamp max_tokens to >= 1 before calling LiteLLM (#523) 2026-02-13 17:08:10 -05:00
qiupinhua 1ae47058d9 fix: refactor code structure for improved readability and maintainability 2026-02-13 18:51:30 +08:00
Re-bin 39dd7feb28 resolve conflicts with main and adapt MiniMax 2026-02-10 16:27:10 +00:00
Re-bin fba5345d20 fix: pass api_key directly to litellm for more robust auth 2026-02-10 02:09:31 +00:00
Xubin RenandGitHub 8af98004b3 Merge pull request #225 from chaowu2009/main 
Drop unsupported parameters for providers.
 2026-02-09 03:52:52 +08:00
Re-bin eb2fbf80da fix: use config key to detect provider, prevent api_base misidentifying as vLLM 2026-02-08 19:31:25 +00:00
Re-bin 2931694eb8 fix: preserve reasoning_content in conversation history for thinking models 2026-02-08 18:37:41 +00:00
Re-bin 299d8b33b3 refactor: replace provider if-elif chains with declarative registry 2026-02-08 07:29:31 +00:00
Re-bin 3b61ae4fff fix: skip provider prefix rules for vLLM/OpenRouter/AiHubMix endpoints 2026-02-08 04:29:51 +00:00
Vincent Wu 3c8eadffed feat: add MiniMax provider support via LiteLLM 2026-02-08 03:55:24 +08:00
Re-bin 572eab8237 feat: add AiHubMix provider support and refactor provider matching 2026-02-07 08:10:05 +00:00
Re-bin 9a98ab1747 Merge PR #145: fix Zhipu AI API key env var 2026-02-07 07:22:51 +00:00
Re-bin 18ec651b34 Merge PR #46: Add DashScope support 2026-02-07 02:52:40 +00:00
Re-bin 7c2aec99a0 resolve conflicts with main 2026-02-07 02:41:28 +00:00
Xubin RenandGitHub 771c918770 Merge pull request #205 from wcmolin/fix/zhipu-api-key 
[Fix-204]: use correct ZAI_API_KEY for Zhipu/GLM models #204
 2026-02-07 10:20:10 +08:00
cwu d7b72c8f83 Drop unsupported parameters for providers. 2026-02-06 12:24:11 -05:00
Re-bin 2096645ff1 resolve conflicts with main 2026-02-06 16:56:02 +00:00
Re-bin 760a369004 feat: fix API key matching by model name 2026-02-06 08:01:20 +00:00
wcmolin fea4a6bba8 fix: use correct ZAI_API_KEY for Zhipu/GLM models 
LiteLLM's zai provider reads ZAI_API_KEY, not ZHIPUAI_API_KEY.
This fixes authentication errors when using Zhipu/GLM models.
 2026-02-06 15:38:25 +08:00
mengjiechenandClaude Opus 4.6 e680b734b1 feat: add Moonshot provider support 
- Add moonshot to ProvidersConfig schema
- Add MOONSHOT_API_BASE environment variable for custom endpoint
- Handle kimi-k2.5 model temperature restriction (must be 1.0)
- Fix is_vllm detection to exclude moonshot provider

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
 2026-02-06 15:25:58 +08:00
Re-bin 301fba568b refactor: remove redundant env var setting, add DeepSeek to docs 2026-02-05 08:55:41 +00:00
Re-bin ac45630116 resolve conflicts with main 2026-02-05 08:51:11 +00:00
Manus AI a0280a1e4a fix: update Zhipu AI API key env var and improve model prefixing 2026-02-05 03:35:46 -05:00
Xubin RenandGitHub b1bce89d4b Merge pull request #32 from pjperez/main 
fix: Use correct 'zai/' prefix for Zhipu AI models in LiteLLM
 2026-02-04 17:10:13 +08:00
ZJUCQR 8499dbf132 add dashscope support 2026-02-03 16:27:15 +08:00
Re-bin eb20cea668 resolve conflicts with main 2026-02-03 06:16:17 +00:00
popcell 8cde0b3072 fix: correct API key environment variable for vLLM mode 2026-02-03 12:14:14 +08:00
Kyya Wang f23548f296 feat: add DeepSeek provider support 2026-02-03 03:09:13 +00:00
Pedro Perez 43fe83755d fix: Use correct 'zai/' prefix for Zhipu AI models in LiteLLM 
LiteLLM expects the 'zai/' provider prefix for Zhipu AI (Z.ai) models,
not 'zhipu/'. This was causing 'LLM Provider NOT provided' errors when
users configured models like 'glm-4.7' without an explicit prefix.

According to LiteLLM docs, the correct format is:
- model='zai/glm-4.7' (correct)
- NOT model='zhipu/glm-4.7' (incorrect)

This fix ensures auto-prefixed models use the correct 'zai/' format.

Fixes: Error when using Zhipu AI models with shorthand names like 'glm-4.7'
 2026-02-02 22:15:21 +01:00
Re-bin a0950cf6b7 resolve conflicts: keep both Gemini and Zhipu support 2026-02-02 13:00:24 +00:00
Re-bin 8d834d9b10 restore CLI commands, keep default model 2026-02-02 12:30:56 +00:00