Subagent tasks hang indefinitely in high-concurrency environment

[huyusong10]

Description

When spawning multiple subagents simultaneously via the task tool in a high-concurrency environment, some subagents occasionally appear to be running (parent shows spinner) but are actually crashed or cancelled. This causes parent agents to wait indefinitely without receiving any error notification.

Visual Evidence

Parent agent shows task as "running" with active spinner animation, but subagent session is actually stuck/crashed with no further output.

Plugins

opencode-worktree@latest; pencode-notifier@latest;

OpenCode version

1.2.27

Steps to reproduce

1. Create a main agent that spawns 20+ subagents concurrently using the task tool
2. Wait for subagents to complete
3. Observe that some subagents hang indefinitely

Screenshot and/or share link

Operating System

macos 26.3.1

Terminal

Ghostty

[github-actions]

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

• Explore subagent hangs indefinitely with Anthropic Claude Opus 4.6 -- no timeout or recovery #13841: Explore subagent hangs indefinitely — no timeout or recovery
• Tasks/Subagents with Codex / OpenAI are frequently getting stuck with no timeout/retry, which then hangs the session forever #11865: Tasks/Subagents frequently getting stuck with no timeout/retry, hanging the session forever
• Sessions hang indefinitely when Task tool spawns subagents via REST API (opencode serve) #6573: Sessions hang indefinitely when Task tool spawns subagents via REST API
• Intermittent hang: session stays running forever until manual interrupt #14769: Intermittent hang: session stays running forever until manual interrupt
• Task tool needs configurable timeout to prevent indefinite hangs #15072: Task tool needs configurable timeout to prevent indefinite hangs
• 1.2.25 subagent/task calls can hang at 0 messages after createUserMessage schema validation failure #17387: Subagent/task calls can hang at 0 messages after schema validation failure

If your issue is distinct (specifically about high-concurrency / 20+ parallel subagents causing hangs rather than a single subagent hanging), please clarify what makes it different from the above.