-
Notifications
You must be signed in to change notification settings - Fork 2.6k
Add /agent/stop endpoint, make max active agents configurable #5826
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Conversation
DOsinga
left a comment
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I like the route, but less agents is not a good idea
| .get_or_try_init(|| async { | ||
| let manager = Self::new(Some(DEFAULT_MAX_SESSION)).await?; | ||
| let max_sessions = Config::global() | ||
| .get_goose_max_active_agents() |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
this is a singleton so making it depend on a configuration variable only somewhat works
more to the point and I think we discussed this previously, I don't think the agent manager should be a cache and definitely not one with max=10 by default - we still have a rumbling subagent problem and that would mean if you start 10 subagents, the main agent dies.
so adding an agent/stop path is the right way, but the clients should either always handle this or the agents should be properly resumed (now it loads the default provider, no extensions etc)
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
so adding an agent/stop path is the right way, but the clients should either always handle this or the agents should be properly resumed (now it loads the default provider, no extensions etc)
I think #5419 resolves some of this
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
restored default to 100
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I think #5419 resolves some of this
it's definitely on the way there. the next step would be to store the active extensions also and then we can move code from resume_agent to here and make sure that everybody uses the agent manager and then we can kill agents and also have agents restore themselves after a system restart
* main: (48 commits) [fix] generic check for gemini compat (#5842) Add scheduler to diagnostics (#5849) Cors and token (#5850) fix sessions coming back with empty messages (#5841) markdown export from URL (#5830) Next camp refactor live (#5706) Add out of context compaction test via error proxy (#5805) fix: Add backward compatibility for conversationCompacted message type (#5819) Add /agent/stop endpoint, make max active agents configurable (#5826) Handle 404s (#5791) Persist provider name and model config in the session (#5419) Comment out the flaky mcp callers (#5827) Slash commands (#5718) fix: remove setx calls to not permanently edit the windows shell PATH (#5821) fix: Parse maas models for gcp vertex provider (#5816) fix: support Gemini 3's thought signatures (#5806) chore: Add Adrian Cole to Maintainers (#5815) [MCP-UI] Proxy and Better Message Handling (#5487) Release 1.15.0 Document New Window menu in macOS dock (#5811) ...
…5826) Signed-off-by: Sai Karthik <kskarthik@disroot.org>
…5826) Signed-off-by: Blair Allan <Blairallan@icloud.com>
Resolves #5824