"No integrations available"; unable to select a model or connect a provider; models unresponsive #57061

Activity

1. opencode-agent commented on Oct 19, 2026

   

   opencode-agentBot

   on Oct 19, 2026 – with OpenCode Agent

   Contributor

   More actions

   Thanks for the report! Those two messages ("No provider selected" and "No integrations available") are what the TUI shows when the background service returns an empty list of models and integrations. That usually means the service itself got stuck or overloaded. It doesn't mean your Go connection was removed. To work out whether the problem is the local service or the requests to Go, could you share:

   1. The Go model(s) you were using, and the directory you started opencode in (for example a project folder or your home directory).
   2. The output of opencode service status and opencode api get /api/info while it's broken.
   3. Whether it works when you run opencode --standalone (this starts the TUI with its own private server).
   4. While it's broken, whether an opencode process is using a lot of CPU or memory (check with top or htop).
   5. Logs: the lines from ~/.local/share/opencode/log/opencode.log around the time it stopped responding, especially lines with role=server and any errors or warnings. If you can reproduce it, running once with OPENCODE_LOG_LEVEL=DEBUG gives more detail. Please remove API keys, prompts and file contents before posting.

   You can reply here with the details, or edit the issue. If you edit it, please also leave a comment saying you've updated it, so we know when you're done.

   I found some related issues. They're ordered by how likely they are to match, so please check whether one of them already covers your problem:

   • core: global location watcher subscribes to all of $HOME — 2M+ file home dirs starve the event loop (~900% CPU, multi-GB memory), /api/model 503 #55937: same "No provider selected" and empty model picker. There, the service hangs at high CPU and /api/model returns 503 when OpenCode is started in a very large home directory.
   • service: recovery loop spawns unbounded serve processes when service becomes unresponsive #57038: the background service stops answering and the client keeps replacing it.
   • Requests always timeout on opencode cli #56371: requests hang without a response until they time out.
   • Not responding at all... #56185: agents not responding at all (few details so far).
2. opencode-agent

   added

   triaging

   on Oct 19, 2026
3. mcbalaam commented on Oct 19, 2026

   

   mcbalaam

   on Oct 19, 2026

   Author

   More actions

   @opencode-agent I was using DeepSeek 4.1 Flash and Claude DontRememberWhichOne 5.5. Tried starting the server in multiple different directories and all of them don't have any available intergations

   /api/get returns {"version":"2.0.31","pid":280483,"urls":["http://127.0.0.1:49374"],"paths":{"tmp":"/tmp/opencode"},"capabilities":{"persistentPty":true}}

   Launching with the --standalone flag does not help

   No unusual CPU/RAM spikes

   Logs you might wanna look at: timestamp=2026-10-19T12:18:24.585Z level=INFO run=e5762978 message="Sent HTTP response" http.span=83000 role=server http.method=GET http.url=/api/integration http.status=499
4. opencode-agent commented on Oct 19, 2026

   

   opencode-agentBot

   on Oct 19, 2026 – with OpenCode Agent

   Contributor

   More actions

   Thanks, that helps a lot! The service is running and answering (/api/info responds, and CPU and memory use are normal). The useful part is your log line: GET /api/integration waited 83 seconds and then the client gave up (499). Before it lists integrations, the server waits for all plugins, including the built-in ones, to finish starting. Sending a prompt waits for the same thing. So if plugin startup never finishes, /connect stays empty and messages hang without an error. That matches what you're seeing. It also happens with --standalone and in every directory, so the cause is probably global: your global config, a saved credential, or a network request that never gets an answer.

   To find out which part gets stuck, could you share:

   1. How you connected Go: did you sign in to OpenCode Console, or paste a Go API key? At startup, a Console connection requests your Console config from opencode.ai. If that request never gets an answer, plugin startup can hang.
   2. Warnings and errors from the server log for one run. Start it with OPENCODE_LOG_LEVEL=DEBUG opencode --standalone, wait about 30 seconds, quit, then run:

      grep 'role=server' ~/.local/share/opencode/log/opencode.log | grep -Ei 'level=(WARN|ERROR)|plugin|activat|config' | tail -n 80

      Please remove any keys or tokens before posting.
   3. Whether this request returns or hangs, and how long it takes:

      time opencode api get /api/integration > /dev/null

   4. Whether these respond from your machine:

      curl -sS -m 20 -o /dev/null -w '%{http_code} %{time_total}s\n' https://opencode.ai
      curl -sS -m 20 -o /dev/null -w '%{http_code} %{time_total}s\n' https://models.dev/api.json

   5. Any plugins entries in ~/.config/opencode/opencode.json(c) and any files in ~/.config/opencode/plugins/, including leftovers from v1 (OpenCode 1.x). Two plugins with the same ID can make plugin startup hang forever (v2: duplicate plugin ID hangs the server — activation defect never opens readiness latch #44975).

   Related issues, ordered by likelihood:

   • v2: duplicate plugin ID hangs the server — activation defect never opens readiness latch #44975: plugin startup never finishes (duplicate plugin ID), so everything that waits for plugins hangs.
   • fix(core): reuse Console config across startup activations #53817: the Console config request delays plugin startup, which holds up models and /connect.
   • core: global location watcher subscribes to all of $HOME — 2M+ file home dirs starve the event loop (~900% CPU, multi-GB memory), /api/model 503 #55937: same "No provider selected" symptom, but caused by an overloaded service. Your CPU and memory are normal, so this is less likely.
5. mcbalaam commented on Oct 20, 2026

   

   mcbalaam

   on Oct 20, 2026

   Author

   More actions

   @opencode-agent models.dev is available via direct pinging, but timeouts within OpenCode as per logs
6. neriousy commented on Oct 21, 2026

   

   neriousy

   on Oct 21, 2026

   Member

   More actions

   Do you still have this problem?

   Try opencode service restart / share your logs
7. opencode-agent

   closed this as not plannedon Oct 22, 2026
8. opencode-agent commented on Oct 22, 2026

   

   opencode-agentBot

   on Oct 22, 2026 – with OpenCode Agent

   Contributor

   More actions

   Closing this for now because we haven't heard back. Thanks again for the details you shared! Your log line, where GET /api/integration waited 83 seconds and then failed with 499, suggests plugin startup never finishes, and prompts wait on the same step. We still need a bit more to find out which part gets stuck.

   If this still happens, please reply here with what I asked for above:

   • whether you connected Go by signing in to OpenCode Console or with an API key
   • the warnings and errors with role=server from one OPENCODE_LOG_LEVEL=DEBUG opencode --standalone run
   • how long opencode api get /api/integration takes, and whether it finishes
   • the two curl checks against opencode.ai and models.dev
   • any plugins entries in your config, or files in ~/.config/opencode/plugins/

   We can reopen the issue once we have them. #44975 and #53817 describe similar ways plugin startup can get stuck, if you'd like to follow them.
9. opencodejr

   reopened this on Nov 3, 2026
10. neriousy

    closed this as not plannedon Nov 12, 2026