Conversation Compaction Failure: Unexpected Error When Attempting to Reduce Context Length
Resolved 💬 16 comments Opened Aug 1, 2025 by Electro-Jam Closed Jan 22, 2026
Bug Description
I restarted a conversation and then I asked the system to reneable a set of flags and work forward from what would be next for this effort and then I got the message:\
\
Context low · Run /compact to compact & continue\
\
So I asked it to /compact \
\
Then go this:\
Error: Error during compaction: Error: Conversation too long. Press esc to go up a few messages and try again.\
\
Not sure what to do from here, but I would not expect this type of response from asking the system to do a /compact .. expecially when the system asked me to do so.
Environment Info
- Platform: win32
- Terminal: vscode
- Version: 1.0.65
- Feedback ID: 902f5297-5edb-49b1-91c1-cb1e8fff33d7
Errors
[{"error":"Error: Tool exit_plan_mode not found\n at nvB (file:///C:/Users/direc/AppData/Roaming/npm/node_modules/@anthropic-ai/claude-code/cli.js:2626:3618)\n at XR (file:///C:/Users/direc/AppData/Roaming/npm/node_modules/@anthropic-ai/claude-code/cli.js:65:19859)\n at Nb (file:///C:/Users/direc/AppData/Roaming/npm/node_modules/@anthropic-ai/claude-code/cli.js:67:42848)\n at $b (file:///C:/Users/direc/AppData/Roaming/npm/node_modules/@anthropic-ai/claude-code/cli.js:67:38369)\n at eS (file:///C:/Users/direc/AppData/Roaming/npm/node_modules/@anthropic-ai/claude-code/cli.js:67:38297)\n at jR (file:///C:/Users/direc/AppData/Roaming/npm/node_modules/@anthropic-ai/claude-code/cli.js:67:38151)\n at LK (file:///C:/Users/direc/AppData/Roaming/npm/node_modules/@anthropic-ai/claude-code/cli.js:67:35214)\n at T8 (file:///C:/Users/direc/AppData/Roaming/npm/node_modules/@anthropic-ai/claude-code/cli.js:65:4841)\n at U3 (file:///C:/Users/direc/AppData/Roaming/npm/node_modules/@anthropic-ai/claude-code/cli.js:67:32459)\n at A.Q.updateContainer (file:///C:/Users/direc/AppData/Roaming/npm/node_modules/@anthropic-ai/claude-code/cli.js:71:1837)","timestamp":"2025-08-01T01:07:53.335Z"},{"error":"Error: Tool exit_plan_mode not found\n at nvB (file:///C:/Users/direc/AppData/Roaming/npm/node_modules/@anthropic-ai/claude-code/cli.js:2626:3618)\n at XR (file:///C:/Users/direc/AppData/Roaming/npm/node_modules/@anthropic-ai/claude-code/cli.js:65:19859)\n at Nb (file:///C:/Users/direc/AppData/Roaming/npm/node_modules/@anthropic-ai/claude-code/cli.js:67:42848)\n at $b (file:///C:/Users/direc/AppData/Roaming/npm/node_modules/@anthropic-ai/claude-code/cli.js:67:38369)\n at eS (file:///C:/Users/direc/AppData/Roaming/npm/node_modules/@anthropic-ai/claude-code/cli.js:67:38297)\n at jR (file:///C:/Users/direc/AppData/Roaming/npm/node_modules/@anthropic-ai/claude-code/cli.js:67:38151)\n at LK (file:///C:/Users/direc/AppData/Roaming/npm/node_modules/@anthropic-ai/claude-code/cli.js:67:35214)\n at T8 (file:///C:/Users/direc/AppData/Roaming/npm/node_modules/@anthropic-ai/claude-code/cli.js:65:4841)\n at U3 (file:///C:/Users/direc/AppData/Roaming/npm/node_modules/@anthropic-ai/claude-code/cli.js:67:32459)\n at A.Q.updateContainer (file:///C:/Users/direc/AppData/Roaming/npm/node_modules/@anthropic-ai/claude-code/cli.js:71:1837)","timestamp":"2025-08-01T01:07:53.384Z"},{"error":"Error: Tool exit_plan_mode not found\n at nvB (file:///C:/Users/direc/AppData/Roaming/npm/node_modules/@anthropic-ai/claude-code/cli.js:2626:3618)\n at XR (file:///C:/Users/direc/AppData/Roaming/npm/node_modules/@anthropic-ai/claude-code/cli.js:65:19859)\n at Nb (file:///C:/Users/direc/AppData/Roaming/npm/node_modules/@anthropic-ai/claude-code/cli.js:67:42848)\n at $b (file:///C:/Users/direc/AppData/Roaming/npm/node_modules/@anthropic-ai/claude-code/cli.js:67:38369)\n at eS (file:///C:/Users/direc/AppData/Roaming/npm/node_modules/@anthropic-ai/claude-code/cli.js:67:38297)\n at jR (file:///C:/Users/direc/AppData/Roaming/npm/node_modules/@anthropic-ai/claude-code/cli.js:67:38151)\n at LK (file:///C:/Users/direc/AppData/Roaming/npm/node_modules/@anthropic-ai/claude-code/cli.js:67:35214)\n at T8 (file:///C:/Users/direc/AppData/Roaming/npm/node_modules/@anthropic-ai/claude-code/cli.js:65:4841)\n at U3 (file:///C:/Users/direc/AppData/Roaming/npm/node_modules/@anthropic-ai/claude-code/cli.js:67:32459)\n at A.Q.updateContainer (file:///C:/Users/direc/AppData/Roaming/npm/node_modules/@anthropic-ai/claude-code/cli.js:71:1837)","timestamp":"2025-08-01T01:07:53.481Z"},{"error":"Error: Language not supported while highlighting code, falling back to markdown: \n at VX (file:///C:/Users/direc/AppData/Roaming/npm/node_modules/@anthropic-ai/claude-code/cli.js:1576:1184)\n at file:///C:/Users/direc/AppData/Roaming/npm/node_modules/@anthropic-ai/claude-code/cli.js:1576:912\n at Array.map (<anonymous>)\n at KX (file:///C:/Users/direc/AppData/Roaming/npm/node_modules/@anthropic-ai/claude-code/cli.js:1576:903)\n at Mx1 (file:///C:/Users/direc/AppData/Ro
Note: Error logs were truncated.
16 Comments
Found 3 possible duplicate issues:
If your issue is a duplicate, please close it and 👍 the existing issue instead.
🤖 Generated with Claude Code
I get same issue. This is super annoying. like stop using claude code annoying imm even using subagents
I lost all work for few hours in same way - all files reverted and all new files lost. I stop also using cursor till this issue is fixed - pointless !!
It seems to be a bug. If going back a single query does not help, try exiting Claude, resuming the conversation, and running compact again.
Ok - thanks
AB Solutions Timisoara
ing.Bedeleu Alexandru
+40 728-442.399
@.***
On Fri, Sep 12, 2025 at 11:18 AM Cenny @.***> wrote:
This issue has been inactive for 30 days. If the issue is still occurring, please comment to let us know. Otherwise, this issue will be automatically closed in 30 days for housekeeping purposes.
Still a problem
What is happening for me is that I am having long-running sessions and the files for going back to an old session get to large and I get this message:
<local-command-stderr>Error: Error during compaction: Error: Conversation too long. Press esc twice to go up a few messages and try again.</local-command-stderr>
This happens both when i'm using the console with --continue for a couple of days and, for the same reason, when I use the CLI streaming interface. It seems that, fo rme on Windows, once the files for history get to long and i've sent "too large" of messages I get stuck in this state.
Further note: I do get this usually just before all of this starts to fail:
"Prompt is too long"
Sometimes I can use a /compact and keep going
Then at some point I'll get the above error.
Please let me know if there are other logs I can get for you to help debug and/or fix this. To me, this is always bad because i'm in the middle of doing work and now I have to do a full session restart and I loose all of the session and that takes a while to get back to the proper conversation / context to keep working.
I wanted to note that I have hit this now 3 times today.
For me this is a major workflow issue for recovering - would be happy to share any info that is needed to fix this.
here is an example:
Here is what I had another agent session go and look at so that you have it for reference.
SESSION FILES (Non-Agent)
| File | Size | Modified | Last Accessed |
|------|------|----------|---------------|
|
2646b07b-*.jsonl| 3.2KB | 2025-11-28 | 2025-12-13 ||
cb53e680-*.jsonl| 18.3MB | 2025-11-29 | 2025-12-13 ||
669cc5d7-*.jsonl| 10.8MB | 2025-11-29 | 2025-12-13 ||
7b63926e-*.jsonl| 2.8MB | 2025-11-29 | 2025-12-13 ||
827fd434-*.jsonl| 1.9MB | 2025-11-29 | 2025-12-13 ||
6a9223bd-*.jsonl| 1.6MB | 2025-11-29 | 2025-12-13 ||
6ce2967a-*.jsonl| 1.1MB | 2025-11-29 | 2025-12-13 ||
387352a6-*.jsonl| 1.3MB | 2025-11-30 | 2025-12-13 ||
2108f6b3-*.jsonl| 114.9KB | 2025-11-30 | 2025-12-13 ||
91fd8467-*.jsonl| 2.3MB | 2025-12-02 | 2025-12-13 ||
6102996b-*.jsonl| 2.3MB | 2025-12-03 | 2025-12-13 ||
c04498ef-*.jsonl| 6.6MB | 2025-12-06 | 2025-12-13 ||
9e7cb7f2-*.jsonl| 1.9MB | 2025-12-06 | 2025-12-13 ||
65d158cf-*.jsonl| 4.4MB | 2025-12-10 | 2025-12-13 ||
593c2d88-*.jsonl| 520KB | 2025-12-10 | 2025-12-13 ||
739138b3-*.jsonl| 1.8KB | 2025-12-10 | 2025-12-13 ||
458dd3c9-*.jsonl| 4.3MB | 2025-12-11 | 2025-12-13 ||
c9d3e31c-*.jsonl| 4.9MB | 2025-12-11 | 2025-12-13 ||
cc9b1169-*.jsonl| 10.2MB | 2025-12-13 | 2025-12-13 ||
526391b8-*.jsonl| 4.9MB | 2025-12-13 | 2025-12-13 ||
8842322f-*.jsonl| 2.8MB | 2025-12-13 | 2025-12-13 ||
14a8d698-*.jsonl| 3.2MB | 2025-12-13 | 2025-12-13 |Oldest file:
agent-fe927874.jsonlfrom Nov 27, 2025Largest files (candidates for deletion):
cb53e680-*.jsonl- 18.3MB (Nov 29)669cc5d7-*.jsonl- 10.8MB (Nov 29)cc9b1169-*.jsonl- 10.2MB (Dec 13 - recent!)Recommended cleanup:
After having the agent run the cleanup:
BEFORE vs AFTER CLEANUP
| Metric | Before | After | Change |
|--------|--------|-------|--------|
| Total Size | 128MB | 36MB | -92MB (72% reduction) |
| Agent Files | 215 | 0 | -215 files |
| Session Files | 36 | 9 | -27 files |
| Empty Files | 14 | 0 | -14 files |
| Total Files | 251 | 9 | -242 files |
---
WHAT WAS DELETED
| Category | Count | Size Freed |
|----------|-------|------------|
| Agent files (subagent histories) | 215 | ~43MB |
| Empty session files (0 bytes) | 14 | 0MB |
| Old sessions (before Dec 10) | 13 | ~49MB |
| TOTAL DELETED | 242 | ~92MB |
---
REMAINING FILES (9 sessions)
| File | Size | Date |
|------|------|------|
|
739138b3-*.jsonl| 1.9KB | Dec 10 ||
593c2d88-*.jsonl| 520KB | Dec 10 ||
65d158cf-*.jsonl| 4.5MB | Dec 10 ||
458dd3c9-*.jsonl| 4.4MB | Dec 11 ||
c9d3e31c-*.jsonl| 5.0MB | Dec 11 ||
14a8d698-*.jsonl| 3.2MB | Dec 13 ||
526391b8-*.jsonl| 4.9MB | Dec 13 ||
8842322f-*.jsonl| 2.9MB | Dec 13 ||
cc9b1169-*.jsonl| 11MB | Dec 13 |All recent work (Dec 10-13) is preserved. The "too long" error should be resolved now.
When I restarted the session, via the --continue I got this message:
"Prompt is too long"
I then ran the /compact
The compact was successful.
Somehow the older logs or larger agent file logs are causing issues with the system over time. This issue is now the 3rd time today and I wanted to give some more info here to see if it would be of help.
another possible note for the team:
I am using the CLI streaming and it looks like I was able to send "to long" of a single message through the CLI interface, once this happened the system got stuck in the state where it would reply after I sent a message and say "message to long" (something close to that). The only way was to delete the current session file -- and yes lost work this time -- but I think it may be, at least in my case, an issue when sending to long of a prompt via the CLI streaming interface.
Having an issue with this too
This issue has been automatically closed due to 60 days of inactivity. If you're still experiencing this issue, please open a new issue with updated information.
This issue was closed incorrectly despite recent human comments. This behavior of the bot is reported at https://github.com/anthropics/claude-code/issues/16497. Please upvote that issue, so maybe it gets noticed.
This issue has been automatically locked since it was closed and has not had any activity for 7 days. If you're experiencing a similar issue, please file a new issue and reference this one if it's relevant.