Skip to content

Request failed with 400 timeout #139

Closed

Description

@CarlduPreez-SA

HI. I am not sure if this has come up before but I am getting the error below when using squad.

Sorry, your request failed. Please try again.

Copilot Request id: 32fbc068-9fc2-4e4a-b0ee-7c84534a5516

GH Request Id: CDD6:3551FD:19F6444:1CBAFA3:699C8280

Reason: Request Failed: 400 {"type":"error","error":{"type":"invalid_request_error","message":"request body is not valid JSON: read tcp 10.125.11.62:2206-\u003e10.125.22.0:42899: i/o timeout"}}

My investigation has led to it being related to the custom agent. If not, I do apologise but wondering if I can get this to work better Its happening constantly on my project after using it for a while. If I start a new project it works fine.

Activity

  1. bradygaster commented on Feb 24, 2026

    @bradygaster
    Collaborator

    I think this is due to the custom agent but we're quickly making improvements and changes and we have a new update you can try coming out very soon. I'll post a reply here when you can upgrade again.

  2. bradygaster commented on Feb 27, 2026

    @bradygaster
    Collaborator

    Response posted with context overflow explanation, immediate workarounds (fresh sessions, upgrade), and note on team's in-flight history compression improvements.

  3. bradygaster commented on Feb 27, 2026

    @bradygaster
    Collaborator

    Hi @CarlduPreez-SA — this is context window overflow, and I can help you work around it today.

    What's Happening

    That 400 timeout error with "i/o timeout" in the body means the request body itself exceeded the API size limit. Here's why:

    As your session accumulates Squad work, the context grows:

    • Agent execution history (every spawn writes to decisions.md)
    • Orchestration logs (launch tables, routing decisions)
    • Team state snapshots (agent charters, skill files, casting assignments)

    This is all context that helps the coordinator and agents understand what's already been done. But it's context, and context has a size budget. On a long session, you can hit that budget.

    Concrete Workarounds — Today

    1. Start a fresh Copilot chat session periodically

    This is the immediate fix. A new session gets a clean context buffer with room to work. I recommend:

    • One session per project sprint (or per ~2 hours of active Squad work)
    • Before starting big orchestration work (refactoring, multi-phase rollouts)
    • Anytime you hit a timeout error

    You don't lose any team state — decisions.md and all agent charters stay in your repo.

    2. Upgrade to the latest version

    Run this command:

    npx github:bradygaster/squad upgrade

    Recent versions include improvements to context management and history compression. While not a complete fix yet, they buy you more session length before hitting the limit.

    Long-Term Solution in Flight

    The team has been actively working on history summarization and context compaction. This means future versions will fit more work into the same context budget. You're not stuck managing sessions forever — this is a known issue with a fix in progress.

    Bottom Line

    Fresh session = immediate relief today. Upgrade = more headroom. And this limitation gets better with each release.

    Sorry for the friction — it's exactly the kind of platform friction we're designed to eliminate.

    — McManus

  4. added a commit that references this issue on Feb 27, 2026
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Projects

    No projects

      Milestone

      No milestone

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions