v0.105.8

Agenta-AI/agentav0.105.8Jul 22, 2026by ashrafchowdury

AI Summary

This release focuses on frontend performance improvements, including painting the playground from disk to reduce boot render costs, and introduces new context metering features.

Key Highlights

Fixed session-mounts-store test call sites
Show full agent-run errors instead of clamping them
Improved drive loading performance with skeletons and retry logic
Added ambient context-usage meter sourced from the model catalog

New Features

Context token-budget indicator in agent composer

Full Release Notes

## What's Changed
* fe-fix(frontend): fix session-mounts-store test call sites by @ardaerzin in https://github.com/Agenta-AI/agenta/pull/5432
* [fix] Show short agent-run errors in full instead of clamping them away by @ashrafchowdury in https://github.com/Agenta-AI/agenta/pull/5395
* [fix] Sandbox draft issue for new agents by @ashrafchowdury in https://github.com/Agenta-AI/agenta/pull/5408
* fix(frontend): drive loading skeletons, non-blocking resolve, retry by @ardaerzin in https://github.com/Agenta-AI/agenta/pull/5430
* [perf] Paint the playground from disk on reload; cut boot render + request cost by @ardaerzin in https://github.com/Agenta-AI/agenta/pull/5433
* [feat] Ambient context-usage meter sourced from the model catalog by @ashrafchowdury in https://github.com/Agenta-AI/agenta/pull/5434
* feat(frontend): context token-budget indicator in agent composer by @ardaerzin in https://github.com/Agenta-AI/agenta/pull/5402
* [release] v0.105.8 by @github-actions[bot] in https://github.com/Agenta-AI/agenta/pull/5431


**Full Changelog**: https://github.com/Agenta-AI/agenta/compare/v0.105.7...v0.105.8