[Bug] Plan subagent incomplete output without tool invocations

I'm encountering an issue suddenly this morning as well where subagents aren't working - immediately showing "Done" with 0 tools called... both with OOTB agents and custom ones.

Found 1 possible duplicate issue:

1. https://github.com/anthropics/claude-code/issues/6529

This issue will be automatically closed as a duplicate in 3 days.

• If your issue is a duplicate, please close it and 👍 the existing issue instead
• To prevent auto-closure, add a comment or 👎 this comment

🤖 Generated with Claude Code

this also happens on Windows

same, task delegation fails, immediately after Claude invokes an agent: Done (0 tool uses · 0 tokens · 0.4s). When asked to debug:

I'm getting this error:

API Error: 400 {"type":"error","error":{"type":"invalid_request_error","message":"tools: Tool names must be unique."},"request_id":"req_1CU0xie1fDPQpSARZnUH6xZh"}

I'm trying to delegate using the Task tool like this:

<invoke name="Task">
<parameter name="subagent_type">general-purpose</parameter>
<parameter name="description">Fix test errors and verify</parameter>
<parameter name="prompt">You need to fix test failures in the package after memory leak fixes were applied.

**Context:**
Three main code changes were completed. Please update test files.

**Your Task:**
1. Run `pnpm check` to see all errors
2. Fix all test files...
</parameter>
</invoke>

Claude Code v2.0.30, M1 MacOS 26.0.1

Seems to be related to #10708

tested without MCP servers and I still have this issue

None of my subagents are working

• Affects: MCP or no MCP subagents (basic tools like grep, read, write, etc.)

Workaround

• Downgrade to v2.0.28

Same, I even renamed everything to be sure lol. I can confirm the downgrade to v2.0.28 works.

Diagnosis

  You've found a bug in Claude Code. Here's what I've discovered:

  1. Your custom agents are correctly configured - The markdown files in ~/.claude/agents/ have proper YAML frontmatter and are recognized by the system
  2. The error is in Claude Code's agent launcher - When trying to instantiate your custom agents, Claude Code is creating duplicate tool names internally, causing the API
  to reject the request
  3. This is not your fault - Your configuration follows the correct format

  The Issue

  The error "tools: Tool names must be unique" is happening at the API level before your agent even starts. Claude Code is likely duplicating tool names when:
  - Merging inherited tools from the parent context
  - Setting up the agent's tool environment
  - Possibly conflicting with MCP tools (I can see you have mcp__ide__* tools active)

  What You Should Do

  1. Report this bug to the Claude Code team at https://github.com/anthropics/claude-code/issues with:
    - The error message
    - Your agent configuration files (which are correctly formatted)
    - That it happens with subagent_type= .....

None of my subagents are working Affects: MCP or no MCP subagents (basic tools like grep, read, write, etc.) Workaround Downgrade to v2.0.28

FYI, still have this bug after downgrading although it's hit or miss when I open a new cc instance whether agents will work.

> None of my subagents are working > > Affects: MCP or no MCP subagents (basic tools like grep, read, write, etc.) > > Workaround > > Downgrade to v2.0.28 FYI, still have this bug after downgrading although it's hit or miss when I open a new cc instance whether agents will work.

What I've noticed happen is that CC will update again every time I open a new terminal and startup CC. Since a fix is sometime "SoonTM", when I open a new terminal:

1. npm install -g @anthropic-ai/claude-code@2.0.28
2. claude --version expect 2.0.28

For now I just keep the terminal with CC open.

I haven't been seeing this on v2.0.37

This issue has been inactive for 30 days. If the issue is still occurring, please comment to let us know. Otherwise, this issue will be automatically closed in 30 days for housekeeping purposes.

This issue has been automatically closed due to 60 days of inactivity. If you're still experiencing this issue, please open a new issue with updated information.

This issue has been automatically locked since it was closed and has not had any activity for 7 days. If you're experiencing a similar issue, please file a new issue and reference this one if it's relevant.