[BUG] Remote trigger runs stuck at "Incubating..." indefinitely — no execution, no error

Preflight Checklist

• [x] I have searched existing issues and this hasn't been reported yet
• [x] This is a single bug report (please file separate reports for different bugs)
• [x] I am using the latest version of Claude Code

What's Wrong?

Environment

• Claude Code (web, claude.ai/code)
• Trigger ID: trig_014AULaWiGr6uDB3PhUgLXPJ
• Environment: env_014w8SxJQuDtHV7Nk25EVVEq (Default / anthropic_cloud)
• Date: April 15, 2026
• No incidents reported on status.anthropic.com at time of filing

What happened

A scheduled remote trigger that was working correctly on April 13–14 began failing on April 15. Two distinct failure
modes appeared in sequence:

Failure 1 — "Resumed session" with no context (7:02 AM and 10:33 AM runs)
The runs appeared to complete (green checkmark in the run list) but opened with "Resumed session" and the agent
responded with "I don't have enough context to help. Could you share: What you tried to run? Any error messages? What
you expected to happen?" — as if the trigger prompt was never delivered. Only 1 step was taken before the agent stopped.

Failure 2 — Stuck at "Incubating..." indefinitely (11:31 AM and later manual runs)
After updating the trigger prompt to a shorter version, subsequent manual runs showed "Incubating..." in the run
transcript and never progressed to actual execution. The run list showed a green checkmark (implying the run was
accepted), but the agent never started. Multiple retries produced the same result.

What I expected
The agent should start executing within a few minutes of being triggered, search the web, and deliver results via Gmail
and Slack MCP connectors.

What actually happened
Runs either started with no prompt context (Failure 1) or never left the "Incubating..." state (Failure 2). No email or
Slack delivery occurred for any April 15 run. The Slack channel and Gmail connectors were verified working
independently.

Run history (all April 15)

┌───────────────┬───────────┬─────────────────────────────────────────┐
│ Time │ Type │ Result │
├───────────────┼───────────┼─────────────────────────────────────────┤
│ 7:02 AM PDT │ Scheduled │ "Resumed session" — no prompt delivered │
├───────────────┼───────────┼─────────────────────────────────────────┤
│ 10:33 AM PDT │ Manual │ "Resumed session" — no prompt delivered │
├───────────────┼───────────┼─────────────────────────────────────────┤
│ 11:31 AM PDT │ Manual │ Stuck at "Incubating..." │
├───────────────┼───────────┼─────────────────────────────────────────┤
│ ~11:45 AM PDT │ Manual │ Stuck at "Incubating..." │
└───────────────┴───────────┴─────────────────────────────────────────┘

Runs that worked (for comparison)

• April 13 at 10:04 AM — Manual ✓
• April 14 at 7:01 AM — Scheduled ✓
• April 14 at 1:47 PM and 2:09 PM — Manual ✓

Additional context

• Trigger uses MCP connectors: Gmail + Slack
• Model: claude-sonnet-4-6
• Allowed tools: WebSearch, WebFetch, Bash
• persist_session: false
• The prompt was updated several times today but the "Incubating..." issue persisted across all prompt versions

What Should Happen?

Agent should run and deliver to Slack and Gmail

Error Messages/Logs

Steps to Reproduce

Steps to Reproduce

For Failure 1 ("Resumed session" / no prompt delivered):

1. Create a remote trigger via the Claude Code scheduled triggers API with a prompt in

job_config.ccr.events[0].data.message.content

1. Update the trigger prompt one or more times via the API (PATCH/update)
2. Allow the trigger to fire on its cron schedule (or manually trigger via "Run now")
3. Open the run transcript — the session opens as "Resumed session" with no memory of the trigger prompt
4. The agent responds asking for context rather than executing the task

For Failure 2 (stuck at "Incubating..."):

1. Using the same trigger from above (or after updating the prompt), click "Run now" from the trigger detail page at

claude.ai/code/scheduled/<trigger_id>

1. Open the newly created run from the run list
2. Observe the run displays only * Incubating... and never progresses
3. Wait 10–15 minutes — the run remains stuck with no tool calls, no output, and no error message
4. Repeated "Run now" attempts produce the same result

Notes:

• Failure 1 appeared to be triggered by updating the prompt close to or after the scheduled fire time
• Failure 2 appeared after multiple manual run attempts in a short window on the same day
• Both failures show a green checkmark in the run list, giving a false impression of success

Claude Model

Sonnet (default)

Is this a regression?

Yes, this worked in a previous version

Last Working Version

last confirmed working run was April 14 at 2:09 PM PDT

Claude Code Version

2.1.109 (Claude Code)

Platform

Anthropic API

Operating System

macOS

Terminal/Shell

VS Code integrated terminal

Additional Information

_No response_