Skip to content

Conversation

@azmy60
Copy link
Collaborator

@azmy60 azmy60 commented Oct 27, 2025

Closes #25

A metadata button that shows token count and used model

image

An auto-compact text that shows when the context limit is <= 15%

image image

A new option to enable/disable auto-compact

image

New mock models for testing (in dev mode)

image

TODOs:

  • Implement context overflow detection
  • Implement auto-compacting feature
    • Better compacting prompt
  • Add disable auto-compacting option to settings page

@azmy60 azmy60 added this to the v2.1 milestone Jan 13, 2026
azmy60 and others added 5 commits January 15, 2026 20:04
Replace provider-based compact generation with prompt-based approach.
Compacting now sends a compact prompt as a regular message, allowing
the conversation to handle it naturally with retries and error recovery.

Key changes:
- Remove generateCompact() from providers, use compact prompt instead
- Add pending message queue to send messages after compacting completes
- Show collapsible compact results with status indicators
- Display auto-compact notice when context threshold approaches
- Rename reducingContext to compacting for clarity
- Add contextOverflow and contextLeftUntilAutoCompact getters
- Remove contextWindow store (consolidated into chat store)

Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>
@azmy60 azmy60 marked this pull request as ready for review January 19, 2026 18:04
azmy60 and others added 14 commits January 20, 2026 01:13
- Use onError handler to track compaction failures instead of try-catch
- Fix retry() to avoid double regeneration during compaction errors
- Extract finishCompaction() helper for cleaner code flow
- Improve context limit calculation to reserve 25% for models <128K tokens
- Comment out expose property to fix vue-tsc compatibility issue

Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>
@azmy60 azmy60 merged commit 4dbfb67 into main Jan 22, 2026
1 check passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

Summarize conversation on certain tokens limit

2 participants