Add something like claude code hooks

[Yazington]

Something like Claude Code Hooks:
https://docs.anthropic.com/en/docs/claude-code/hooks

Would be nice to chck which files its modifying and give a small message saying things like don't modify this specific function or use countLines to get the number of lines in this file.

It would be managed by the user, they provide implementation in the case the hook "activates". The user would provide a callback that shows a message like You have crossed 100 lines, here is the file structure and how the project is organized, lets rethink this specific file and create a seperate module for this domain.

We can also do something code <-> natural language where a sub agent/sub task is created to do the changes independantly from the main task at hand, but give a tiny context saying changes were made for this (this might not be possible in parallel, but if we can rethink this, it would be great).

[abhisek]

I was about to create a new discussion and found this, so adding as a comment to bump this request.

I am the developer of https://github.com/safedep/gryph

We are leveraging AI coding agent's hook feature to log and maintain an audit trail for security and auditing purposes in a local sqlite database.

So far we have leveraged the hooks feature of Claude Code, Gemini CLI, Cursor, OpenCode to log agent events.

I wanted to add support for Kilo code as well but going through the docs, I did not find a hook or a plugin interface. My needs are, ability to call a local CLI command for:

• Before/After Tool Call
• Before/After filesystem operations (read/write)
• Before/After command execution

Is this something that is already possible or in roadmap?