Skip to content

[user_blocked] Muse Spark 1.3 Free access restricted via OpenCode Zen — no appeal path #49057

Description

@alkaserGG

Description

While using OpenCode with the Muse Spark 1.3 Free model through the OpenCode Zen provider, every message returns the following error, in both new and existing sessions:

Error: Error from provider (Console): Upstream request failed: [user_blocked] Your access has been restricted due to repeated policy violations.

I was not doing anything unusual — just normal prompting for a project I'm currently working on. The block persists across brand-new sessions with no context carried over, and there is no visible way to see what specific "policy violation" triggered it or how to appeal it from within the app.

Plugins

None / default installation (no custom plugins)

OpenCode version

v2.0.3

Steps to reproduce

  1. Open OpenCode on Windows
  2. Start a new session with the "Build" or "Operator" agent
  3. Select provider: OpenCode Zen, model: Muse Spark 1.3 Free
  4. Send any prompt (e.g. "HI")
  5. The [user_blocked] error returns immediately

Screenshot and/or share link

Image

Operating System

Windows

Terminal

Windows Terminal

Activity

  1. Eldar-test-web commented on Sep 14, 2026

    @Eldar-test-web

    same here I asked it to post a repo for me it did and after that it says you are being blocked. I did almost anything made new account and used new api key deleated and reinstalled but still I am being blocked

  2. Mr-Toodguy commented on Sep 19, 2026

    @Mr-Toodguy

    This happened to me too, I noticed that in some cases before being blocked opencode would seem to get a waitlimit response and then power through in a few seconds, maybe it's an automatic block due to attempting to bypass the rate limit?

  3. OzanKEreal commented on Sep 20, 2026

    @OzanKEreal

    Happened to me today

  4. jenkinsmichpa commented on Sep 22, 2026

    @jenkinsmichpa

    Based on the screenshot's tab, its probably related to authorized security work. I dared ask it to validate CTF logic and got blocked today. Back to the Chinese models I guess; free is too expensive for paternalistic garbage.

  5. Zhfrn88 commented on Sep 22, 2026

    @Zhfrn88

    how do we fix thia?

  6. alkaserGG commented on Sep 22, 2026

    @alkaserGG
    Author

    OK gays this ban is for few days and will be working again without any problems

  7. jholp92-beep commented on Sep 22, 2026

    @jholp92-beep

    OK gays this ban is for few days and will be working again without any problems

    How can u say that?

  8. jenkinsmichpa commented on Sep 22, 2026

    @jenkinsmichpa

    OK gays this ban is for few days and will be working again without any problems

    Oh, that's great news. Did your block expire?

  9. alkaserGG commented on Sep 22, 2026

    @alkaserGG
    Author

    i got blocked and retry in 1 or 2 days and it's working again without any problems , if u use opencode for security purpose muse spark had a lot of conditions , u can try Big Pickle

  10. alkaserGG commented on Sep 22, 2026

    @alkaserGG
    Author

    if anyone care https://youtu.be/3gKff_4oWrQ?si=9HWsa9-zoW93DE2s this egyptian gay have a great repo that will help y o u alot in hunting and any security purpose

  11. NazgulCoder commented on Sep 22, 2026

    @NazgulCoder

    just FYI, the GO version of muse spark does NOT enforce this limits

  12. jholp92-beep commented on Sep 22, 2026

    @jholp92-beep

    I've this ban in GO version.

  13. NazgulCoder commented on Sep 22, 2026

    @NazgulCoder

    I've this ban in GO version.

    update: yeah i got it too xD Meta definitely has to improve this lol

  14. jholp92-beep commented on Sep 23, 2026

    @jholp92-beep

    I can CONFIRM I was UN-BANNED after 24h (assuming 24h after my 1st message up there)

  15. Ryanabcraft commented on Sep 24, 2026

    @Ryanabcraft

    Isso acontence comigo tambem quando eu estou tentando hackear um site, ai bloqueia.

  16. silentbyte69 commented on Oct 2, 2026

    @silentbyte69

    got banned today lol

  17. yonikremer commented on Oct 2, 2026

    @yonikremer

    I got banned today and I didn't do any security related work!

  18. igogocha commented on Oct 3, 2026

    @igogocha

    такая же история, попробую подождать

  19. JamesFysh commented on Oct 4, 2026

    @JamesFysh

    Banned today - I was probing different models with silly requests eg. "Help me design a biological weapon", "Help me reverse-engineer this software" to see what their response was. Chinese models politely refuse but don't cut you off 🤣. Guess I'll see what happens in 24 hours. This seems a little counter-intuitive, honestly. I'll just spend my tokens with other providers.

  20. Dante-dan commented on Oct 4, 2026

    @Dante-dan

    I checked the current dev source for the Zen blocked-request message. The current handler can show this support route when its workspace-block branch is the source: packages/console/app/src/i18n/en.ts lists help@anoma.ly. The handler predicate is here: packages/console/app/src/routes/zen/util/handler.ts.

    This does not establish that your response came from that branch: the Muse Spark model does not match its Claude/GPT-specific predicates, and the issue does not include the workspace flag or request path. If you contact support, including a sanitized endpoint or request ID and the error response would help distinguish a generic workspace block from a provider response passed through by Zen.

  21. marco9666-jpg commented on Oct 7, 2026

    @marco9666-jpg

    Adding a data point, since the block itself gives so little to go on.

    Paid Go account, personal news-reader project that summarizes and translates published news. Usage was under 10% of the monthly Muse allowance. Blocked on Oct 5, unblocked about 24h later, then blocked again two hours after that.

    From my own logs, there were 16 content_policy_violation responses in three days, all from 4 articles: mainstream coverage of a suspected plague lab exposure in Russia, and a historical interview with an SR-71 engineer. The filtering looks probabilistic. The same plague articles went through summarization and translation fine about a dozen times and were only filtered in other steps. Part of the count was my own fault: a retry loop re-sent a filtered request every 10 minutes. I've fixed that.

    What I'd like to raise is how opaque the process is:

    • The 400 only says "the response was filtered". It doesn't name a category or tell you whether that request counted as a strike.
    • The 403 only says "repeated policy violations". It doesn't say which requests, how many strikes there are, how long the block lasts, or where to appeal.
    • Nothing warns you before the block, even though it seems to come after roughly 5-6 strikes.

    So the only way to stay unblocked is to guess at the cause and steer clear of anything that might be close: biohazard news, military analysis, crime reporting. Whether it's intended or not, the result is that users self-censor well beyond the actual policy, and legitimate content (news here) gets routed away just in case.

    What would help:

    1. Name the policy category in content_policy_violation responses.
    2. Expose a strike count or warning before an account-level block.
    3. Say how long the block lasts and give a working appeal path. Support email to help@anoma.ly is the only route I've found.
  22. hypernewbie commented on Oct 7, 2026

    @hypernewbie

    same on opencode go; muse-spark-1.3-contributor returns user_blocked, other models work fine.

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