Compaction bug - prompt token limit #10385

Activity

1. Genrie9

   added

   bugSomething isn't workingSomething isn't working

   on Jan 24, 2026
2. github-actions

   assigned

   rekram1-node

   on Jan 24, 2026
3. github-actions

   added

   windows

   on Jan 24, 2026
4. github-actions commented on Jan 24, 2026

   

   github-actionsBot

   on Jan 24, 2026 – with GitHub Actions

   Contributor

   More actions

   This issue might be a duplicate of existing issues. Please check:

   • /compact doesn't utilize prompt caching #10342: /compact doesn't utilize prompt caching
   • Feature Request: Custom compaction threshold to trigger earlier #10017: Feature Request: Custom compaction threshold to trigger earlier
   • Feature Request: Configurable context limit and auto-compaction threshold #8140: Feature Request: Configurable context limit and auto-compaction threshold
   • Compaction Failing to Run on Time #6286: Compaction Failing to Run on Time
   • [BUG]: Auto compaction not triggering reliably with Anthropic models #6068: [BUG]: Auto compaction not triggering reliably with Anthropic models
   • Auto-compaction enabled by default, but context_length_exceeded errors still occur in agent workflows #8089: Auto-compaction enabled by default, but context_length_exceeded errors still occur in agent workflows
   • config not work(feature compact) #9349: config not work(feature compact)

   These issues all involve compaction-related problems with prompt token limits and context management. Your issue specifically mentions using Opus 4.5 with a context fill followed by running /compact. Issues #6286 and #6068 describe similar symptoms where compaction fails with Opus models when context is nearly full, and issue #9349 reports auto-compaction failing to work properly even when configured.

   Feel free to ignore if your specific case is distinct from these.
5. chelebyy commented on Feb 15, 2026

   

   chelebyy

   on Feb 15, 2026

   More actions

   I am experiencing this exact issue on Windows 11.

   Crucial Update: I've discovered that this happens even in a completely empty directory. When I send a simple "hello", the system still tries to send ~377k tokens, specifically with a massive amount of tool input (~217k tokens).

   This confirms that the bug isn't necessarily about the project size, but rather OpenCode incorrectly scanning global/system paths or failing to clear session context on Windows.

   I've tested this with Kimi, GLM, and MiniMax (Free/Paid) and the behavior is identical across all providers.

   

   👍React with 👍1Reacted by juzweb
6. rekram1-node commented on Feb 15, 2026

   

   rekram1-node

   on Feb 15, 2026

   Contributor

   More actions

   @chelebyy You are using a plugin for Omo, try using without that plugin, also feel free to share opencode debug config, I imagine you have a bunch of things being sent that are bloating context.
7. rekram1-node

   closed this as duplicateon Feb 15, 2026
8. chelebyy commented on Feb 15, 2026

   

   chelebyy

   on Feb 15, 2026

   Last edited by chelebyy

   More actions

   @rekram1-node yes the problem is Omo
   debug_config.txt

   `