Repository navigation
Replies: 1 comment
|
This also breaks my workflow, in https://github.com/cslib-community/cslib-frontier , which builds the idea of separate worker and coordinator agents. If they happen to be run by the same account, they can no longer communicate via watching. Perhaps we can just add a boolean parameter to watch to specify whether to include comments from the authenticated user. I imagine it's still useful to have it off in some circumstances to avoid an extra wake-up, but overall, having it on seems better to me. |
0 replies
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.
Note
Created by Codex (GPT-6.1 Sol) on behalf of @YLfjuk, who reviewed and approved this proposal.
My agents post PR comments through my GitHub account and identify themselves in the comment text. Sometimes I reply directly on GitHub with feedback or instructions.
The current PR watch excludes comments from the authenticated account. That prevents the agent from waking on its own comments, but also excludes my human-written replies because both have the same GitHub author.
I'd like a way for my replies to wake the watching agent while continuing to suppress its own comments and avoid reply loops.
One possible approach would be an agreed marker identifying agent-written comments. Another could be tracking the IDs of comments posted by the agent. These are suggestions rather than a preference for a particular implementation.
Related: #35293 proposes filtering comment authors. This request concerns distinguishing human and agent comments within the same account.
Would this be a direction you'd consider?
All reactions