Cron job burning tokens, how to best manage

1. We have 2 cron job files, why?

~/.clawdbot/cron/jobs.json - Main Clawdbot gateway config (visible via clawdbot cron list)
/Users/felixkrause/clawd/cron/jobs.json - Workspace-specific config (hidden from CLI)
The travel agent cron (01fd2ad3-5d07-477e-8984-359f5b07ffb9) is in the workspace config, not visible via clawdbot cron list.

why do we have 2? can we easily merge them? I want all my cron jobs to be visible through the CLI

2. Travel Agent Context Issue

• Currently runs every 5 minutes, session persists (555KB session file)
• Each run loads: SOUL.md, HEARTBEAT.md, last-heartbeat-messages.json, trip notes + Claude reasoning (~6K tokens)
• Token usage: ~3.5M tokens/day
What we changed now:

• Changed schedule to every 4 hours
• Need a way to clean up the session file after each run, or not to keep it in the first place

3. All Cron Jobs Should Not Store Context

Currently crons seem to persist sessions that accumulate over time.

Desired behavior: Every cron job should spawn a fresh session, do its work, and completely clean up - no persistent context between runs.

Questions for Clawdbot:

1. Is there a deleteSessionAfterRun option (separate from deleteAfterRun which deletes the cron job)?
2. Can we configure all crons to use ephemeral sessions by default?
3. Can we consolidate all cron configs into one file visible via CLI?
Was this page helpful?