fix(cloud-agent-next): collect outcome health on the legacy 3-minute cron

[eshurakov]

Problem

After #6906 deployed, Cloudflare kept firing the previous */3 trigger while the new scheduled() handler matched only */5. Every tick logged Cloud Agent scheduled handler received an unrecognized cron and ran no collection, so no agent_execution/agent_open rows were emitted and the pre-existing nested cloud_agent_outcome_aggregate stream stopped (~13:34Z).

Evidence from production: every scheduled invocation from 13:45Z through 14:06Z carries Event.Cron = "*/3 * * * *", with no invocation at the */5-only slots; the Cloudflare dashboard and the wrangler deploy log both show */5, and re-adding the trigger in the dashboard did not change the firing schedule.

Fix

Accept both expressions in scheduled(). The outcome window already floors controller.scheduledTime to 5-minute buckets, so an extra tick only re-emits a bucket, which the dashboard dedupes by windowStart.

This makes a stale trigger unable to silently disable collection. It does not resolve the underlying trigger/scheduler mismatch, which still needs Cloudflare-side reconciliation.

Checks

• src/server.test.ts: 105 passed
• pnpm run lint: 0 warnings, 0 errors
• pnpm run format:check: clean
• pnpm run typecheck (tsgo): clean

[kilo-code-bot]

Code Review Summary

Status: No Issues Found | Recommendation: Merge

Executive Summary

The change adds a OUTCOME_AGGREGATE_CRONS set so the scheduled handler also accepts the stale */3 trigger; the extra ticks only re-emit a 5-minute outcome bucket that the window already floors, and I found no correctness, security, or memory-leak issues in the changed lines.

Files Reviewed (2 files)

• services/cloud-agent-next/src/server.ts
• services/cloud-agent-next/src/server.test.ts

---

_{Reviewed by deepseek-v4.1-flash · Input: 0 · Output: 0 · Cached: 0}

_{Review guidance: REVIEW.md from base branch main}