Replies: 1 comment 1 reply
-
|
I've got something similar built out here (I think, based on what I am understanding at least) as a full Compiler-IR, micro-opcode encoding, structured transport protocol that compresses what is being sent into an ultra-compact form and then decompresses the output. Also includes a delta transport layer that updates only the changes rather than the entire file. These two projects would work extremely well together, as they tackle different problems, but in a very complementary way. |
Beta Was this translation helpful? Give feedback.
1 reply
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Uh oh!
There was an error while loading. Please reload this page.
Uh oh!
There was an error while loading. Please reload this page.
-
Hey guys,
I am currently thinking about adding some functionality to the MCP which would allow directly manipulating the graph nodes, which are connected to your code files. Through that you would also shape the code itself on your filesystem, effectively making CBM a complete read/write infrastructure on top of your code. This likely would safe some roundtrips and input tokens, as the agent will only need to understand the dependencies of the code through the graph and will then directly only write on those nodes which he wants to change (which then through the database translate over to your actual code files async in the background).
What do you think of this? It is currently kinda only a thought experiment of mine. Would really appreciate your opinions :)
Beta Was this translation helpful? Give feedback.
All reactions