Owner

ModelEngine-Group

8 indexed repositories · View on GitHub

nexent
Nexent is a zero-code platform for auto-generating production-grade AI agents using Harness Engineering principles — unified tools, skill...

Python · 5874 stars
fit-framework
FIT: 企业级AI开发框架，提供多语言函数引擎（FIT）、流式编排引擎（WaterFlow）及Java生态的LangChain替代方案（FEL）。原生/Spring双模运行，支持插件热插拔与智能聚散部署，无缝统一大模型与业务系统。

Java · 2117 stars
app-platform
AppPlatform 是一个前沿的大模型应用工程，旨在通过集成的声明式编程和低代码配置工具，简化和优化大模型的训练与推理应用的开发过程。本工程为软件工程师和产品经理提供一个强大的、可扩展的环境，以支持从概念到部署的全流程 AI 应用开发。

Java · 1457 stars
DataMate
DataMate is an enterprise-level data processing platform designed for model fine-tuning and RAG retrieval.

TypeScript · 368 stars
unified-cache-management
Persist and reuse KV Cache to speedup your LLM.

C++ · 334 stars
flexai

C++ · 233 stars
pipeline
构建部署完整 ModelEngine 的脚本仓库。

PLpgSQL · 22 stars
elsa

JavaScript · 13 stars

Beginner friendly Everything

Open All

默认职责提示词教的终止方式与 CodeAct 守卫判定相矛盾，问答场景高频陷入重试循环直至 max_steps Open

Difficulty 4/5 3-5 days Newbie friendliness 55/100

ModelEngine-Group/nexent#3986 · Opened Sep 21, 2026
[Feature]: SGLang Release 接入 UCM CI/CD Open

feature request

Difficulty 5/5 Over a week Newbie friendliness 35/100

ModelEngine-Group/unified-cache-management#1416 · Opened Sep 21, 2026
[Release]: Release checklist for v0.9.0 Open

Difficulty 5/5 Over a week Newbie friendliness 20/100

ModelEngine-Group/unified-cache-management#1408 · Opened Sep 19, 2026
[Feature]: Lookup-triggered first-shard prefetch for CacheStore Open

feature request performance

Difficulty 5/5 Over a week Newbie friendliness 38/100

ModelEngine-Group/unified-cache-management#1382 · Opened Sep 16, 2026
[Feature]: NUMA-aware rank-partitioned shared CacheBuffer Open

feature request performance

Difficulty 5/5 Over a week Newbie friendliness 35/100

ModelEngine-Group/unified-cache-management#1381 · Opened Sep 16, 2026
fix(mcp): unbind deleted MCP tools from agent drafts Open

Improvement Medium

ModelEngine-Group/nexent#3935 · 1 assignee · Opened Sep 16, 2026
[Feature] MiniMax M3 适配 Open

Difficulty 5/5 Over a week Newbie friendliness 20/100

ModelEngine-Group/unified-cache-management#1378 · Opened Sep 15, 2026
[Feature]: Posix Store 多后端探测与高可用 Open

documentation

Difficulty 5/5 Over a week Newbie friendliness 20/100

ModelEngine-Group/unified-cache-management#1376 · Opened Sep 15, 2026
[Feature]: UCM Store integration with HiCache for Hybrid models Open

feature request

Difficulty 5/5 Over a week Newbie friendliness 25/100

ModelEngine-Group/unified-cache-management#1373 · Opened Sep 15, 2026
[Feature]: Support multimodal input kv cache Open

feature request

ModelEngine-Group/unified-cache-management#1370 · 1 assignee · Opened Sep 14, 2026
OpenAIModel: default observer is the class, not an instance — TypeError on any call that omits it Open

Difficulty 2/5 1-3 hours Newbie friendliness 86/100

ModelEngine-Group/nexent#3921 · Opened Sep 14, 2026
[Request] Optional endpoint-evidence example before selecting an external A2A collaborator Open

Difficulty 3/5 1-2 days Newbie friendliness 58/100

ModelEngine-Group/nexent#3916 · Opened Sep 13, 2026
[Release]: Release checklist for v0.8.0rc1 Open

Difficulty 5/5 Over a week Newbie friendliness 20/100

ModelEngine-Group/unified-cache-management#1363 · Opened Sep 11, 2026
UCM release: blocked upstream capabilities Open

Difficulty 5/5 Over a week Newbie friendliness 15/100

ModelEngine-Group/unified-cache-management#1360 · Opened Sep 11, 2026
[Feature]: Support Minimax-M3 with UCM on vLLM and vLLM-Ascend Open

feature request

Difficulty 5/5 Over a week Newbie friendliness 30/100

ModelEngine-Group/unified-cache-management#1358 · Opened Sep 11, 2026
[Feature]: Support Kimi-K3 with UCM on vLLM and vLLM-Ascend Open

feature request

Difficulty 5/5 Over a week Newbie friendliness 25/100

ModelEngine-Group/unified-cache-management#1357 · Opened Sep 11, 2026
[Feature]: Support DeepSeek-V4-Pro-0813 with UCM on vLLM and vLLM-Ascend Open

feature request

Difficulty 5/5 Over a week Newbie friendliness 35/100

ModelEngine-Group/unified-cache-management#1356 · Opened Sep 11, 2026
[Feature]: Support DeepSeek-V4-Flash-0731 with UCM on vLLM and vLLM-Ascend Open

feature request

Difficulty 5/5 Over a week Newbie friendliness 35/100

ModelEngine-Group/unified-cache-management#1355 · Opened Sep 11, 2026
[Feature]: Support GLM-5.3-Flash with UCM on vLLM and vLLM-Ascend Open

feature request

Difficulty 5/5 Over a week Newbie friendliness 35/100

ModelEngine-Group/unified-cache-management#1354 · 1 comment · Opened Sep 11, 2026
[Bug]: In PD decoupled deployment, using toy_proxy_server to send proxy requests will trigger an error when the `min_tokens` parameter is added, and the proxy service cannot process requests normally. Open

bug

Difficulty 2/5 1-3 hours Newbie friendliness 30/100

ModelEngine-Group/unified-cache-management#1348 · Opened Sep 9, 2026
fix(k8s): correct shared storage PV rendering Open

Difficulty 3/5 1-2 days Newbie friendliness 72/100

ModelEngine-Group/nexent#3869 · Opened Sep 7, 2026
fix(deploy): include official skills in K8s offline packages Open

Difficulty 3/5 1-2 days Newbie friendliness 76/100

ModelEngine-Group/nexent#3868 · Opened Sep 7, 2026
UCM 2026 Q4 RoadMap Open

Difficulty 5/5 Over a week Newbie friendliness 25/100

ModelEngine-Group/unified-cache-management#1314 · Opened Sep 1, 2026
GlobalThreadPool singleton: unsynchronised `__new__` and silently ignored `max_workers` Open

Difficulty 3/5 1-2 days Newbie friendliness 68/100

ModelEngine-Group/nexent#3820 · Opened Aug 30, 2026
SERVICE_ROLE_KEY silently falls back to SUPABASE_KEY — privilege boundary erased Open

Difficulty 3/5 1-2 days Newbie friendliness 68/100

ModelEngine-Group/nexent#3819 · Opened Aug 30, 2026
`convert_string_to_list` silently drops negative integers and non-digit entries Open

Difficulty 2/5 1-3 hours Newbie friendliness 78/100

ModelEngine-Group/nexent#3818 · Opened Aug 30, 2026
Improvement: `sort_models_by_id` only sorts by the **first** character of the id Open

Difficulty 2/5 1-3 hours Newbie friendliness 84/100

ModelEngine-Group/nexent#3817 · Opened Aug 30, 2026
Praise: tasteful streaming-thinking-token handling in `_process_thinking_tokens` Open

Difficulty 5/5 Over a week Newbie friendliness 15/100

ModelEngine-Group/nexent#3815 · Opened Aug 30, 2026
`TenantConfigManager.update_single_config` has no `value` argument — it can only bump timestamps Open

Difficulty 3/5 1-2 days Newbie friendliness 68/100

ModelEngine-Group/nexent#3814 · Opened Aug 30, 2026
[Bug] `calculate_term_weights` divides by zero when input is all stop words Open

Difficulty 1/5 Under an hour Newbie friendliness 88/100

ModelEngine-Group/nexent#3813 · Opened Aug 30, 2026
Question: what's the intended lifecycle for `AgentRunManager._conversation_run_counts`? Open

Difficulty 5/5 Over a week Newbie friendliness 35/100

ModelEngine-Group/nexent#3812 · Opened Aug 30, 2026
tokenizer: `position_factor` mixes character-length with token-index Open

Difficulty 2/5 1-3 hours Newbie friendliness 78/100

ModelEngine-Group/nexent#3811 · Opened Aug 30, 2026
`get_model_name_from_config` raises `KeyError` instead of returning empty string Open

Difficulty 2/5 1-3 hours Newbie friendliness 84/100

ModelEngine-Group/nexent#3810 · Opened Aug 30, 2026
[Request] Add Serply as a web search tool option Open

Difficulty 3/5 1-2 days Newbie friendliness 35/100

ModelEngine-Group/nexent#3748 · Opened Aug 22, 2026
[Feature]: Performance Optimization of the Multi-Rank Shared-Cache Load Path Open

feature request performance

ModelEngine-Group/unified-cache-management#1265 · 1 assignee · Opened Aug 21, 2026
Clarification on CacheStore buffer validation (-50000 too small buffer) for non-MLA models and the hard-coded 1024 floor Open

Difficulty 4/5 3-5 days Newbie friendliness 42/100

ModelEngine-Group/unified-cache-management#1261 · Opened Aug 20, 2026
[Doc]: add new docs website Open

documentation

Difficulty 5/5 Over a week Newbie friendliness 25/100

ModelEngine-Group/unified-cache-management#1258 · Opened Aug 20, 2026
Module name typo: `model_managment_app.py` should be `model_management_app.py` Open

Difficulty 2/5 1-3 hours Newbie friendliness 88/100

ModelEngine-Group/nexent#3698 · Opened Aug 18, 2026
[Improvement] 后端接口支持分页查询 Open

ModelEngine-Group/nexent#3691 · 1 assignee · Opened Aug 18, 2026
`remove_think_blocks` regex deletes content that is not in a `<think>` block Open

Difficulty 2/5 1-3 hours Newbie friendliness 78/100

ModelEngine-Group/nexent#3689 · Opened Aug 17, 2026
[Document] 文档网站增加全局搜索功能 Open

Difficulty 4/5 3-5 days Newbie friendliness 35/100

ModelEngine-Group/nexent#3660 · Opened Aug 13, 2026
统一 MCP 连通性检测与模型配置的交互层级 Open

Difficulty 4/5 3-5 days Newbie friendliness 50/100

ModelEngine-Group/nexent#3659 · Opened Aug 13, 2026
知识库概览总结支持后台运行及任务队列状态监控 Open

Difficulty 5/5 Over a week Newbie friendliness 35/100

ModelEngine-Group/nexent#3658 · Opened Aug 13, 2026
知识库构建支持 ZIP/文件夹拖拽及自动识别可用文件 Open

Difficulty 4/5 3-5 days Newbie friendliness 48/100

ModelEngine-Group/nexent#3657 · Opened Aug 13, 2026
[Bug] tool_db.py add_tool_field: NoneType object has no attribute 'params' when tool.params is NULL Open

Difficulty 2/5 1-3 hours Newbie friendliness 82/100

ModelEngine-Group/nexent#3650 · Opened Aug 12, 2026
[Draft][Feature]: UCM inference-sim tool (Toolkit sub-issue) Open

Difficulty 5/5 Over a week Newbie friendliness 25/100

ModelEngine-Group/unified-cache-management#1225 · Opened Aug 12, 2026
智能体版本管理的时候，版本只能通过按钮一个一个选择删除，不方便 Open

Difficulty 3/5 1-2 days Newbie friendliness 55/100

ModelEngine-Group/nexent#3646 · Opened Aug 12, 2026
[Feature][ToolKit]: 前缀重复率统计工具（字符级/token级/block级） Open

Difficulty 5/5 Over a week Newbie friendliness 35/100

ModelEngine-Group/unified-cache-management#1221 · 1 reaction · Opened Aug 12, 2026
修改skill的时候，用户组做变更增加一个用户组，显示提交技能失败 Open

Difficulty 3/5 1-2 days Newbie friendliness 45/100

ModelEngine-Group/nexent#3645 · Opened Aug 12, 2026
[Feature][ToolKit]: KV hash-directory recovery tool (recreate shards without restart) Open

Difficulty 5/5 Over a week Newbie friendliness 45/100

ModelEngine-Group/unified-cache-management#1220 · Opened Aug 12, 2026

Showing the newest 100

This page lists what was indexed most recently. Advanced filter has the whole inventory, narrowed by language, difficulty and how long a task takes.

Open advanced filter

ModelEngine-Group

Showing the newest 100

Get new issues in your inbox