[Q&A]

[khalilammarr]

What kind of help do you need?

Integrations or tooling

Your question

I Think somehow integrating tihs repo , then installing the new update of antigravity 1.20.5 will cause a prompt error [Idea]

Environment and context

AntiGravity 1.20.5

What have you already tried?

Reinstalling

Before posting

• I checked the README, FAQ, and relevant docs first.

[sickn33]

Thanks for reporting this. This sounds similar to the known truncation/context-overflow issue that can happen when an IDE/integration loads too many skills into one prompt at startup, but I can’t confirm that yet from the current details.

Could you share:

1. the exact error text or a screenshot of the prompt error,
2. your OS,
3. whether you installed the full library or only a small subset of skills,
4. whether this is happening in the IntelliJ/IDEA integration specifically.

As a quick workaround, please try loading only a few skills instead of the full library, or use the activation script to keep only the bundles you need active at once.

If you are on Windows and Antigravity keeps reopening into the same broken session, please also check the truncation recovery steps here:

• README: Windows truncation/context crash loop
• docs/users/windows-truncation-recovery.md

If you can post the exact error, I can tell you whether this is the same truncation issue or something new in 1.20.5.

[khalilammarr]

Thanks for the response. The OS is Linux , the error is the following : Agent terminated due to error
You can prompt the model to try again or start a new conversation if the error persists. Error
Unknown: Agent execution terminated due to error.

Trajectory ID: 2f45c473-eb96-4399-b406-260d3a629bf6
Error: agent executor error: could not convert a single message before hitting truncation
(1) attached stack trace
  -- stack trace:
  | google3/third_party/jetski/cortex/cortex.(*CascadeManager).executeHelper.func1
  | 	third_party/jetski/cortex/cascade_manager.go:2088
  | [...repeated from below...]
Wraps: (2) agent executor error
Wraps: (3) attached stack trace
  -- stack trace:
  | google3/third_party/jetski/cortex/chatconverters/chatconverters.init
  | 	third_party/jetski/cortex/chatconverters/trajectory_chat_converter.go:44
  | runtime.doInit1
  | 	third_party/go/gc/src/runtime/proc.go:8105
  | runtime.doInit
  | 	third_party/go/gc/src/runtime/proc.go:8072
  | runtime.main
  | 	third_party/go/gc/src/runtime/proc.go:258
  | runtime.goexit
  | 	third_party/go/gc/src/runtime/asm_amd64.s:1774
Wraps: (4) could not convert a single message before hitting truncation
Error types: (1) *withstack.withStack (2) *errutil.withPrefix (3) *withstack.withStack (4) *errutil.leafError

And yes i have the full skills loaded , and this is happening at every prompt so i guess its not an overload problem idk, i cant even use antigravity anymore : Here i found some people facing the same problem , for me personally removing the .agent folder in the home directory fixed the issue but i wouldnt use it without the skills honestly

I switched to VsCode with copilot the repo works just fine

[sickn33]

@khalilammarr thanks for posting the full stack trace. That confirms this is the same truncation/context-overflow class of failure, not a separate skill bug.

The detail that removing the .agent folder fixed it is also a strong signal that two things were happening together:

• the full library was too large for that host/integration path
• the host had stored broken local session/state and kept reopening into it

So your workaround makes sense: clearing the local state removed the bad session, but if you restore the full skills tree again, the same class of problem can come back.

For Linux/macOS, the safer pattern is to keep only a focused active set instead of exposing the whole library at once. That is what the overload recovery guide is for:

• docs/users/agent-overload-recovery.md

And for the integration side, the long-term fix is still:

• manifest/index first
• lazy-load only the requested skills
• enforce a hard per-turn cap/token budget

The fact that VS Code + Copilot works fine with the repo also points to this being a host-loading problem rather than bad skill content.

Thanks again — this extra detail was useful, and it helped confirm the Linux side of the same issue family.