- v0ens2026-09-0132993
本帖使用社区开源推广,符合推广要求。我申明并遵循社区要求的以下内容:
- 我的帖子已经打上 开源推广 标签: 是
- 我的开源项目完整开源,无未开源部分: 是
- 我的开源项目已链接认可 LINUX DO 社区: 是
- 我帖子内的项目介绍,AI生成、润色内容部分已截图发出: 是
- 以上选择我承诺是永久有效的,接受社区和佬友监督: 是
以下为项目介绍正文内容,AI生成、润色内容已使用截图方式发出
来源: 什么?ACE强制收费了,不怕!我们有《干翻A狗》【测试版上线】【已开源】
现已正式开源,用法详见:
github.com

GitHub - oce-ai/oce: OpenContextEngine is a self-hosted, ACE-compatible...
OpenContextEngine is a self-hosted, ACE-compatible code retrieval service. It indexes source files with cAST-aware chunking, stores metadata in PostgreSQL or SQLite, performs dense vector retrieval in Milvus 3.0, and reranks results with an LLM before coverage-aware selection.
- 刚刚开源很多边界估计没有覆盖到 也有不少功能需要继续完善,后续我将会写一个详细的文档
- 建议使用原仓库匹配的客户端
uv tool install "opencontextengine-client[mcp]"-
ace-tool-rs这些虽然兼容,但是有些地方处理还是有些不同的也希望佬们多多提
pr、issues下一步计划
- 开发 dsh、pi 插件与agent对话过程深度集成实现
- git 提交索引(据我所知 Augment Code 之前是有这个功能的)
部分评分
已经做了几个中小仓库的评分:
Python项目: flask
OCE Retrieval Evaluation
- Date: 2026-08-25
- Service:
http://127.0.0.1:8986- Repository:
C:\Users\admin\Projects\oce-ai\flask- Repository SHA:
22d924701a6ae2e4cd01e9a15bbaf3946094af65- Queries: 50
- Source blobs considered: 229
- Upload failures: 0
- Peak process RSS: 43.4 MB
- Scoring: Top-1 (1 pt, ranking ceiling) + nDCG@10 (0..1 pt, result usability)
Total
Points / 100 |
Top-1 |
Mean nDCG@10 |
Score % |96.33 |
48 / 50 (96.0%) |
0.967 |
96.3% |Categories
Category |
Top-1 |
Mean nDCG@10 |
Points |api_usage |
4 / 4 |
1.000 |
8.00 / 8 |architecture_understanding |
2 / 3 |
0.873 |
4.62 / 6 |call_chain |
9 / 10 |
0.939 |
18.39 / 20 |component_location |
4 / 4 |
1.000 |
8.00 / 8 |configuration_lookup |
4 / 4 |
1.000 |
8.00 / 8 |cross_language |
3 / 3 |
0.867 |
5.60 / 6 |error_handling |
3 / 3 |
1.000 |
6.00 / 6 |file_exact_match |
3 / 3 |
1.000 |
6.00 / 6 |semantic_feature |
6 / 6 |
0.953 |
11.72 / 12 |symbol_location |
10 / 10 |
1.000 |
20.00 / 20 |混合项目: cc-switch
OCE Retrieval Evaluation
- Date: 2026-08-25
- Service:
http://127.0.0.1:8986- Repository:
C:\Users\admin\Projects\cc-switch- Repository SHA:
40cac1a68edf8c9e7b3a89125cf40bb93a348404- Queries: 50
- Source blobs considered: 1035
- Upload failures: 0
- Peak process RSS: 47.0 MB
- Scoring: Top-1 (1 pt, ranking ceiling) + nDCG@10 (0..1 pt, result usability)
Total
Points / 100 |
Top-1 |
Mean nDCG@10 |
Score % |87.31 |
45 / 50 (90.0%) |
0.846 |
87.3% |Categories
Category |
Top-1 |
Mean nDCG@10 |
Points |api_usage |
3 / 4 |
0.595 |
5.38 / 8 |architecture_understanding |
1 / 3 |
0.710 |
3.13 / 6 |call_chain |
9 / 10 |
0.683 |
15.83 / 20 |component_location |
4 / 4 |
1.000 |
8.00 / 8 |configuration_lookup |
4 / 4 |
1.000 |
8.00 / 8 |cross_language |
3 / 3 |
1.000 |
6.00 / 6 |error_handling |
2 / 3 |
0.667 |
4.00 / 6 |file_exact_match |
3 / 3 |
1.000 |
6.00 / 6 |semantic_feature |
6 / 6 |
0.932 |
11.59 / 12 |symbol_location |
10 / 10 |
0.938 |
19.38 / 20 |- 后续将评测脚本上传至 GitHub
主楼 - Tim2026-09-0113037