💻💣 SQL-aware extraction, cross-language impact edges, and advanced change forecasting?
💻💣 SQL-aware extraction, cross-language impact edges, and advanced change forecasting?
I've seen tools like graphify, codegraph, and others but don't know which, if any, is worth trying.
This is something Claude kinda handles okay, but can loose details on.
I've seen tools like graphify, codegraph, and others but don't know which, if any, is worth trying.
This is something Claude kinda handles okay, but can loose details on.
https://benzi.fly.dev/benchmark
https://benzi.fly.dev/benchmark
comments · posted on 2026.09.10 at 21:22:44 (c=1, p=4)
comments · posted on 2026.09.10 at 21:22:44 (c=1, p=4)
把"文本搜索"换成"图查询":
→ Agent 想要哪个符号/依赖,直接查
→ 只带回需要的节点,不整文件塞进上下文
→ 跨文件跨语言的调用关系,regex 表达不了
官方对比实测:token 成本降最多 36%。
#AgentDev #CodeGraph
把"文本搜索"换成"图查询":
→ Agent 想要哪个符号/依赖,直接查
→ 只带回需要的节点,不整文件塞进上下文
→ 跨文件跨语言的调用关系,regex 表达不了
官方对比实测:token 成本降最多 36%。
#AgentDev #CodeGraph
shaaf.dev/post/migrati...
#SoftwareEngineering #CodeGraph #AI #Coding #Agents #OpenSource #Refactoring #Quarkus #Java
shaaf.dev/post/migrati...
#SoftwareEngineering #CodeGraph #AI #Coding #Agents #OpenSource #Refactoring #Quarkus #Java
MegaMemory 소개
코딩 에이전트를 며칠 이상 쓰다 보면 같은 작업을 반복하게 됩니다. 어제 세 시간 동안 인증 흐름을 읽고 어디를 고쳐야 하는지 파악해 두었는데, 오늘 새 세션을 열면 에이전트는 그 파일들을 처음 보는 것처럼 다시 읽습니다. 이 문제를 푸는 흔한 방향은 코드베이스를 미리 인덱싱해 두는 것이고, PyTorchKR 에도 CodeGraph나 codebase-memory-mcp처럼 코드를 그래프로 만들어 두는 도구가…
MegaMemory 소개
코딩 에이전트를 며칠 이상 쓰다 보면 같은 작업을 반복하게 됩니다. 어제 세 시간 동안 인증 흐름을 읽고 어디를 고쳐야 하는지 파악해 두었는데, 오늘 새 세션을 열면 에이전트는 그 파일들을 처음 보는 것처럼 다시 읽습니다. 이 문제를 푸는 흔한 방향은 코드베이스를 미리 인덱싱해 두는 것이고, PyTorchKR 에도 CodeGraph나 codebase-memory-mcp처럼 코드를 그래프로 만들어 두는 도구가…
You may be right. I have a gut feeling that integrating a RAG pipeline with cases/law addresses the context issue.
This framework handles the context of millions of lines of code quite well with codegraph and RAG. Context is capped at 256k.
You may be right. I have a gut feeling that integrating a RAG pipeline with cases/law addresses the context issue.
This framework handles the context of millions of lines of code quite well with codegraph and RAG. Context is capped at 256k.
TencentDB Agent Memory is a memory-centric framework that builds a multi-agent team with persistent assets (Chat Memory, Skills, Wiki, CodeGraph) via Memory Hub, enabling cross-agent reuse, access control, and layered memory (L0–L3) to reduce rework and accelerate agent workflows. (1/2)
TencentDB Agent Memory is a memory-centric framework that builds a multi-agent team with persistent assets (Chat Memory, Skills, Wiki, CodeGraph) via Memory Hub, enabling cross-agent reuse, access control, and layered memory (L0–L3) to reduce rework and accelerate agent workflows. (1/2)
TencentDB Agent Memory is a team-centered memory platform that captures and organizes conversations, documents, code, and skills into layered assets (Chat Memory, Skills, Wiki, CodeGraph) with access controls, enabling cross-team reuse, faster cold starts, and iterative (1/2)
TencentDB Agent Memory is a team-centered memory platform that captures and organizes conversations, documents, code, and skills into layered assets (Chat Memory, Skills, Wiki, CodeGraph) with access controls, enabling cross-team reuse, faster cold starts, and iterative (1/2)
TencentDB Agent Memory is a modular, team-oriented memory system for AI agents that captures and organizes cross-session knowledge (Chat Memory, Skills, Wiki, CodeGraph) into layered, reusable assets within a Memory Hub, enabling cross-agent sharing, controlled access, and faster (1/2)
TencentDB Agent Memory is a modular, team-oriented memory system for AI agents that captures and organizes cross-session knowledge (Chat Memory, Skills, Wiki, CodeGraph) into layered, reusable assets within a Memory Hub, enabling cross-agent sharing, controlled access, and faster (1/2)
What are your experiences?
Care to share?
What are your experiences?
Care to share?
https://youtu.be/4dVIZqR6FW8
#Use #Clojure #codegraph
https://youtu.be/4dVIZqR6FW8
#Use #Clojure #codegraph