Qwen Code v0.26.0 ships session-targeted skill refresh and managed-agent preflight
qwen-code-review-bot · ghdev · 2026-10-22
Alibaba's open-source coding agent tool Qwen Code released v0.26.0 with a batch of engineering improvements centered on managed agents and serving:
- Managed-agent: session-targeted Skill/Extension refresh, bounded publication guard waits, local and SSH workspace file references, a read-only --check preflight, and a configuration template shipped with the reference distribution
- Serve/runtime: warm session creation for workspace-isolated bound sessions, private CSI file turns routed through Hosted Harness, and containment of hosted stdio MCP servers in a process group
- Fixes & tests: recovery of committed wakes after writer loss, session sizing from the daemon budget, ENOTEMPTY retries in teardowns, and slow-host scheduling budgets
The release is an engineering iteration aimed at multi-session, hosted agent workloads.
More from coding & agent
- Nous Research ships official Hermes Agent plugin for Meshy AI 3D generation via MCP — NousResearch · 2026-10-22
- Support AI Regurgitating Stale Docs Is a Growing Failure Mode — Give It Real Platform View — joshwhiton · 2026-10-22
- CallBaba launches BabaBench, a benchmark testing AI agents on real patient advocacy tasks — mattturck · 2026-10-22
- 70%-95% failure rates: devs say AI agents still too unreliable for real workflows — Luke17man · 2026-10-22
- How do you actually test agents before production? Devs tackle failure-path testing — Last-Bluejay-4443 · 2026-10-22
- A Reddit long-read proposes the Risk and Judgement pattern for bounding agent autonomy — mostly_deterministic · 2026-10-22