Create

Sign in to ReadmeX

or

How we handle your data: Privacy Policy

ReadmeX

Your AI reading companion

What would you like to catch up on?

Pick a briefing, or ask your own question.

Your community and people briefings stay personal to you.

Story

Codex Uses OWL MCP to Diagnose PDF Parsing Failures

AI summary

In a test reported by 掘金 · 人工智能, the author connected Codex CLI to Guance’s OWL MCP and instructed it to query runtime logs for a failed PDF import in a Milvus-backed AI knowledge base. Codex used eight tool calls, including two business-log queries, to identify a PDFBox parsing error; the author then compared the six returned records with the platform and local logs and found the event details consistent.

In-depth insight

The account demonstrates a concrete workflow in which Codex queries centralized observability data through MCP, while the author independently verifies the returned records and retains human review.

上一篇,我把 PDF 导入时散开的异常日志收到了观测云里,一条报错终于能完整地读了。

接下来我想省掉另一个动作:出了问题,先翻日志,再复制一大段给 AI。既然 Codex 已经能帮我读代码,能不能让它自己查运行日志?

这次就拿知识库的 PDF 解析做个实测。我给 Codex 一个文档编号和故障描述,让它去观测云找依据,最后再回控制台逐条核对。

先只告诉它现象

我的项目是一个接了 Milvus 的 AI 知识库。资料先解析成文本,再做切分、向量化和入库,供后面的问答检索使用。一份 PDF 没有进入知识库,单看这个结果,很难知道停在哪一步。

Milvus 负责后面的向量存储与检索,前面的程序还得先把 PDF 读成文本。如果连文件都没读出来,排查就得从解析开始。只说一句“没导进去”,还分不清这些情况。

#Codex#OWL MCP#Guance Observability#DataKit#Milvus#PDFBox

Original · 9 min readRead the original
Group chat

No comments yet. Start the conversation.