[BUG] RemoteTrigger routines dispatch but never execute — last_fired_at updates, last_runs stays empty, skip-cascade on hourly crons

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?

RemoteTrigger-based routines are dispatching on schedule (last_fired_at updates each fire) but the agent runtime never starts — last_runs is completely empty across all triggers. No sessions are created, no tool calls execute, no errors are surfaced.
Additionally, hourly cron triggers show a 3-4 hour gap between last_fired_at and next_run_at, suggesting the scheduler treats the prior (empty) dispatch as still "running" and skips subsequent slots.

Affected triggers

All 6 RemoteTriggers on my account exhibit the same behavior:
| Trigger | Cron | last_fired_at (UTC) | next_run_at (UTC) | last_runs |
|---------|------|-----------------------|---------------------|-------------|
| trig_01K41QsrbsLnnydNBemcSHzj (COS) | 0 0-2,11-23 * * 1-6 | 2026-06-04T22:04:03Z | 2026-06-05T01:05:51Z | empty |
| trig_01BLhhP4eQ5PcyxxL5jghZ6M (Site Tracker) | 0 9-23,0-4 * * * | 2026-06-04T21:09:52Z | 2026-06-05T01:02:53Z | empty |
| trig_019aTpcKp24TCHFCViBx91RZ (CAC Tracker) | 0 11-23,0-4 * * * | 2026-06-04T22:01:39Z | 2026-06-05T01:03:16Z | empty |
| trig_01NKfuyMSkGkD5yxsChvrKrP (Weekly Pull) | 0 13 * * 1 | — | 2026-06-08T13:05:10Z | empty |
| trig_011jUSKCN3b726na6cgFotKn (Weekly Funnel) | 0 15 * * 1 | — | 2026-06-08T15:04:53Z | empty |
| trig_01J3mhKXeTqjJSxousDZ9GfR (Daily Funnel) | 30 14 * * * | — | 2026-06-05T14:35:31Z | empty |

Evidence of dispatch without execution

The API response for each trigger shows:

• enabled: true
• last_fired_at updates on schedule (the scheduler IS dispatching)
• last_runs: [] — zero completed sessions, zero turns
• No ended_reason set
• next_run_at advances, but with multi-hour gaps on hourly crons (expected: ~1 hour gap)

What Should Happen?

Each cron fire should create a session, execute the agent turn with the configured prompt/MCP connections, and populate last_runs with the session result.

Error Messages/Logs

None. That is part of the bug. The API returns no errors, `ended_reason` is an empty string on all triggers, and no failed or stuck sessions appear in the UI. The only way to detect the failure is noticing that `last_runs` is empty despite `last_fired_at` advancing — there is zero user-facing signal that routines have stopped working.

Steps to Reproduce

1. Create a RemoteTrigger via the API with an hourly cron (e.g., 0 11-23,0-4 * * *)
2. Attach MCP connectors (Shopify, Slack, Gmail, etc.) and a valid prompt
3. Confirm trigger shows enabled: true via GET /v1/code/triggers/{id}
4. Wait for the cron to fire
5. GET the trigger again — last_fired_at has advanced, but last_runs is []
6. Wait for the next cron slot — it does not fire on time. next_run_at shows a 3-4 hour gap instead of the expected 1 hour
7. No error is surfaced at any point

Claude Model

Sonnet (default)

Is this a regression?

Yes, this worked in a previous version

Last Working Version

1.11186

Claude Code Version

Claude Code version: 2.1.123

Platform

Anthropic API

Operating System

macOS

Terminal/Shell

Terminal.app (macOS)

Additional Information

_No response_