APScheduler default misfire_grace_time is too strict for long-running Claude commands

[sakhnenkoff]

Problem

The AsyncIOScheduler in JobScheduler.__init__ (src/scheduler/scheduler.py) is created without explicit job_defaults, which means APScheduler uses its default misfire_grace_time=1 second. This is far too strict for a bot where Claude command execution routinely takes 1-10+ minutes.

Any job that can't start within 1 second of its scheduled time is silently skipped.

Proposed Fix

Set misfire_grace_time=None and coalesce=True in job defaults:

self._scheduler = AsyncIOScheduler(
    job_defaults={
        "misfire_grace_time": None,  # Always run, no matter how late
        "coalesce": True,            # Merge multiple missed runs into one
    }
)

• misfire_grace_time=None guarantees every job fires exactly once, regardless of how late
• coalesce=True prevents duplicate execution if multiple runs were missed

This combination means the worst case is one late run, never a skipped one.

I have a working implementation with tests — happy to open a PR.

Environment

• claude-code-telegram v1.6.0
• APScheduler 3.11.x

[chrisgwynne]

I took a pass at this as a small patch.

Patch

misfire_grace_time=1

Optional test

If you want, I can open a PR, but I don't currently have push permission to the repo (and I'm avoiding forks).

[chrisgwynne]

Follow-up (clean): the minimal change is just setting job defaults on the AsyncIOScheduler.

Rationale: long-running Claude executions mean the scheduler can't always start jobs within APScheduler's default 1s misfire grace. Late is better than skipped.

I also drafted an optional unit test asserting if you want it.