Hey folks - so been working with the Amp plug in for a bit and enjoying it - however, not sure if it was something i did or if it’s a bug - but sometimes when i compress the context it clears the context and there doesn’t seem to be a way to get it back - as a result i now get this error and cannot continue in the present thread as the error just repeats with each request.
Error
400 {"type":"error","error":{"type":"invalid_request_error","message":"messages: text content blocks must be non-empty"}}
this unfortunately requires me to create a whole new thread and lose all the context of the old thread ( which sometimes is frustrating given the amount of dialog and clarifications that occur while generating the context for the model to understand the objective - not to mention the context cost for having to create a new thread and have it read all the project files back in again )
so this begs the question - am I doing something wrong by compressing the context and if so, is there a way to restore the context?
respectfully
thanks!!
Hey @ericwelsh.dev Thank you for reporting the issue.
Is this in VS Code or the CLI? What version are you using?
Sorry for the inconvenience.
Hi @PriNova thanks for the quick reply !
Yep its in VS Code:
Version: 1.102.1 (user setup)
Commit: 7adae6a56e34cb64d08899664b814cf620465925
Date: 2025-07-15T16:41:17.147Z
Electron: 35.6.0
ElectronBuildId: 11847422
Chromium: 134.0.6998.205
Node.js: 22.15.1
V8: 13.4.114.21-electron.0
OS: Windows_NT x64 10.0.26100
with the Amp plugin
Amp (Research Preview) v0.0.1753186353
please let me know if you need any additional information 
Hi @PriNova thanks for looking into this - so it happened again ( had to create a new thread and attempt to copy paste what was left of the conversation in hopes it would be able to pick up parts of it of what had been done verses was still had to be complete - that in itself wasn’t a big deal - but rather the amount of “tokens” used to building up the holistically context of the service architecture I’ve built before ( in efforts to provide a better / clearer context for it )
any thoughts? as not compressing the context is consuming an excessive about of tokens (I’ve already had to buy close to $200 just to get back to where I started before, I lost the context last time and I’m not sure I can afford to continue to do that again.)
respectfully
thanks!!
Hey @ericwelsh.dev Thank you for your feedback. The team is aware of the issue and they are working on it to fix asap.
Sorry for the inconvenience.
Hi @PriNova , another note - could be related … but I just lost an entire thread (with context of course too) - was in the middle of an operation and the whole thread blank, and the amp header showed “new thread”, I was like , ok I can just go back and select my old thread … nope - it was gone … the last thread on the list of was from 12 days ago.)
I’m sorry, I’m at a loss here - I’m a student and this is coming out of my pocket, so these kinds of setbacks are not only time consuming, expensive and frustrating. I’m loving the tool but it’s getting to the point where the costs are outweighing the benefits in a sense - not sure if we can refuel my balance or increase the tokens a bit until these issues are resolved at all ?
respectfully
thanks!!