Skip to content

Claude Code v2.1.219 → v2.1.220: nested subagent messages now propagate through the stream #527

Description

@claude-yolo

WHY this matters

If you're building anything that drives Claude Code programmatically — CI runners, orchestration layers, SDK hosts — three changes land together that you need to know about.

🚀 New capabilities

Deep subagent stream forwarding (v2.1.219+)
Nested subagent messages now appear in --output-format stream-json with parent_tool_use_id set to the spawning Agent tool call ID. Before v2.1.219, nested messages were silently dropped. SDK consumers can now rebuild the full nesting tree by following those IDs. Requires --forward-subagent-text or CLAUDE_CODE_FORWARD_SUBAGENT_TEXT.

MCP server error detection in CI (v2.1.219+)
The system/init stream event now includes mcp_server_errors — an array of --mcp-config entries skipped at startup due to bad config. CI can now gate on a non-empty array. Previously: silent skips.

New capability string interrupt_cancel_queued_v1 added to system/init capabilities array.

⚠️ Security fix (v2.1.218)

Subagent hooks in .claude/agents/ markdown files now only fire if the folder has been trusted via the workspace trust dialog. Non-interactive CI sessions: hooks suppressed until you trust the folder.

🔥 Quiet breaking change (v2.1.218)

: is now reserved in subagent names (for plugin-scoped agents like my-plugin:reviewer). Any agent file with : in its name is silently skipped. Silent migration hazard if you have such files.


Source PR: thevibeworks/claude-code-docs#1109

Assign: @lroolle

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions