Chat Session ID and Completion Status API #185849
Replies: 3 comments 1 reply
-
|
Yeah, i think you’re not missing anything, this just isn’t exposed right now Copilot Chat doesn’t give session IDs, completion events, or status info, so theres no clean way to know when an agent is done. That’s why orchestration(agent -> wait -> next agent) ends up relying on hacks like files or timeouts Definitely feels like a missing API feature, especially for multi-agent workflows like yours |
Beta Was this translation helpful? Give feedback.
-
|
You’re not overlooking anything — today the What’s missing todayThere’s currently no supported way to:
As a result, Copilot Chat behaves more like a fire-and-forget UI feature than an orchestratable execution primitive. This makes sequential or multi-agent workflows (agent → validate → next agent) effectively impossible without brittle workarounds like polling files, timeouts, or heuristic output parsing. Impact on extension authorsFor extensions that orchestrate multi-step or multi-agent workflows, this creates a major gap. Other VS Code execution models (tasks, terminals, debug sessions) all provide:
Copilot Chat currently provides none of these, forcing extensions into fragile and unreliable patterns. What would enable reliable orchestrationAny of the following would unlock clean, supported workflows:
At the moment, there’s no clean or supported solution — this is a genuine platform gap rather than a misuse of the API. If multi-agent workflows are a target use case for Copilot-based extensions, exposing agent execution lifecycle signals would be a foundational improvement. |
Beta Was this translation helpful? Give feedback.
-
|
s |
Beta Was this translation helpful? Give feedback.
Uh oh!
There was an error while loading. Please reload this page.
-
Select Topic Area
Question
Copilot Feature Area
Visual Studio
Body
When using vscode.chat. to programmatically invoke Copilot Chat functionality, there's no way to:
Use Case : Our extension orchestrates multi-step and multi-agent workflows
Currently we're forced to use file-based workarounds where agents create "completion marker" files that we poll for, which is fragile and unreliable.
What we need - some way to detect when a specific chat participant has finished responding, whether that's:
This would enable reliable agent orchestration similar to how tasks, terminals, and debug sessions provide completion signals.
Beta Was this translation helpful? Give feedback.
All reactions