@colton_drake I think we're aligned with the trend toward open-weight frontier models, where we're seeing growing demand. Over time, we'd like to support as many stable harnesses and models as possible.
Report
congrats!can developers choose different LLM models for the same harness? how quickly can you switch models without changing the agent configuration?
Saying state restore is best effort and cannot prevent a race condition is rare on a launch day. You mentioned wanting idempotent agent replies in your own layer. Where do you think that line sits between the infra and the product on top, Darren?
@lucasjpols Appreciate the feedback! An analogy I would use is DB backup/restore, there would be no guarantee on a perfect restore with absolute no loss of rows. For most product user cases, I do feel this is sufficient, considering that agent crash should not be a high frequent event. We are still trying to put more common product concerns into our infra layer, while still let developers to keep their own part.
Report
Congrats Darren! The channels part is what sells it for me. I correct my coding agent from the phone mid-run all the time, and that flow is usually held together with duct tape. Curious: when a WhatsApp message lands while the agent is mid-task, does it interrupt the run or queue until the current step finishes?
you may request interrupt + another input message (interrupt first, then handle the new message)
you may just send another input message which will be queued
Report
@darren_y_lu interrupt first is exactly what I need when correcting from the phone and queue covers the rest. thanks for the link, good luck with the rest of the week
Report
The cross-channel part is the most compelling to me. If the same long-running agent can keep context across Slack, Telegram, CLI, and web without each app team rebuilding the plumbing, that removes a lot of hidden operational work. Curious how you think about audit logs for those handoffs over time.
@senlin_wang Thanks! Right now we provide limited activity logs in our system, but we are developing more profound logging support.
Report
The cross-platform memory is the part I’m most curious about. If a user starts debugging a problem through Discord, then continues the conversation through email or a support portal later, does the agent maintain the same context and history, or does each channel create an isolated session?
For developer communities, users rarely stay on one platform, so I’m wondering how agents handle context continuity when conversations move across different channels.
@new_user___176202668afdbb135b4f71f > If a user starts debugging a problem through Discord, then continues the conversation through email or a support portal later,
The conversation will be continuous for AgentSky agents.
what runtime or channel are you most excited to add next? are you prioritizing integrations based on community requests or technical demand?
@colton_drake I think we're aligned with the trend toward open-weight frontier models, where we're seeing growing demand. Over time, we'd like to support as many stable harnesses and models as possible.
congrats! can developers choose different LLM models for the same harness? how quickly can you switch models without changing the agent configuration?
@carter_garcia Thanks!
Yes, for compatible ones.
Switching models is considered changing agent configuration. It takes about ~5 sec to deploy the new change.
Can AgentSky automatically scale when multiple users start interacting with the same agent? how do you handle sudden spikes in agent sessions?
@dylan_russell
multiple users get their own sessions with same agent (same MDs, same skills, same capabilities)
Under the hood, we have an elastic sandbox infra which can handle new agent sandbox requests on demand.
Saying state restore is best effort and cannot prevent a race condition is rare on a launch day. You mentioned wanting idempotent agent replies in your own layer. Where do you think that line sits between the infra and the product on top, Darren?
@lucasjpols Appreciate the feedback! An analogy I would use is DB backup/restore, there would be no guarantee on a perfect restore with absolute no loss of rows. For most product user cases, I do feel this is sufficient, considering that agent crash should not be a high frequent event. We are still trying to put more common product concerns into our infra layer, while still let developers to keep their own part.
Congrats Darren! The channels part is what sells it for me. I correct my coding agent from the phone mid-run all the time, and that flow is usually held together with duct tape. Curious: when a WhatsApp message lands while the agent is mid-task, does it interrupt the run or queue until the current step finishes?
@artem_t Thank you! For now, we support 2 modes, please check out https://agentsky.dev/docs/api
you may request interrupt + another input message (interrupt first, then handle the new message)
you may just send another input message which will be queued
@darren_y_lu interrupt first is exactly what I need when correcting from the phone and queue covers the rest. thanks for the link, good luck with the rest of the week
The cross-channel part is the most compelling to me. If the same long-running agent can keep context across Slack, Telegram, CLI, and web without each app team rebuilding the plumbing, that removes a lot of hidden operational work. Curious how you think about audit logs for those handoffs over time.
@senlin_wang Thanks! Right now we provide limited activity logs in our system, but we are developing more profound logging support.
The cross-platform memory is the part I’m most curious about. If a user starts debugging a problem through Discord, then continues the conversation through email or a support portal later, does the agent maintain the same context and history, or does each channel create an isolated session?
For developer communities, users rarely stay on one platform, so I’m wondering how agents handle context continuity when conversations move across different channels.
@new_user___176202668afdbb135b4f71f
> If a user starts debugging a problem through Discord, then continues the conversation through email or a support portal later,
The conversation will be continuous for AgentSky agents.