Add scheduled agents with cron execution and webhook triggers

[emal-avala]

Summary

• agent schedule subcommand — full CRUD for persistent schedules (add, list, run, remove, enable, disable)
• agent daemon — background cron loop checking schedules every 30 seconds, executing due jobs
• Webhook triggers — optional --webhook-port on daemon starts an HTTP server; external systems POST to /trigger?secret=<key> to invoke schedules on demand
• Per-schedule config — model override, cost limits, turn limits, permission mode
• Session persistence — each run creates a session resumable via /resume <id>
• Cron parser — built-in 5-field cron with *, */N, N-M, N,M support + next_after() for computing next fire time

Example usage

agent schedule add "0 9 * * *" --prompt "run tests and fix failures" --name daily-tests --webhook
agent schedule add "0 */6 * * *" --prompt "check for security updates" --name security-scan --max-cost 0.50
agent schedule list
agent schedule run daily-tests
agent daemon --webhook-port 8090

# External trigger:
curl -X POST http://localhost:8090/trigger?secret=<webhook_secret>

Architecture

~/.config/agent-code/schedules/*.json   ← persisted schedule definitions
agent daemon                            ← checks every 30s, runs matching schedules
  └─ ScheduleExecutor.run_once()        ← creates QueryEngine, runs prompt, saves session
agent daemon --webhook-port 8090        ← also starts Axum webhook server
  └─ POST /trigger?secret=<key>         ← finds schedule by secret, runs it

Test plan

• 18 new unit tests passing (cron parsing, storage CRUD, webhook lookup, serialization)
• All 262 existing tests still passing
• cargo clippy clean (0 warnings)
• cargo fmt --check clean
• CI passes on this PR

[chatgpt-codex-connector]

You have reached your Codex usage limits for code reviews. You can see your limits in the Codex usage dashboard.