# Monitor website availability and technical SEO

Recipe: website-health@1.0.0
Publisher: AgentAction

## Goal
Perform bounded availability checks and daily technical SEO comparisons; preserve evidence, report new problems and verify recovery.

## Runtime
AgentAction recurring workflows with built-in public web checks and workspace email notifications.

## Your inputs
- Page URL: One public HTTPS page per instance; create additional instances for other pages or sites.
  Example: https://example.com/
- Expected content: Optional text required in the returned HTML.
  Example: Welcome

## Runtime requirements
- Use the recurring-agent runtime with persistent state and an owner-approved schedule.
- Configure workspace email recipients for notifications.
- Same-origin redirects are permitted; explicitly approve additional origins before following them.

## Example output (synthetic)
Example (synthetic): page unavailable after two consecutive checks; one critical finding opened; notification queued. Recovery resolves the finding after a successful check.

## Validate with your agent
These are procedures to run in your sandbox, not completed live-agent tests.
1. Outage and recovery: Use a controlled target that returns 503 twice, then 200.
   Expected: One finding and alert, followed by a recovery event.
2. SEO regression: After the baseline, introduce noindex and run the next daily SEO check.
   Expected: A critical finding references the inspected URL and indexing directive.

## Boundaries
- Public HTTPS reads of configured origins only; no credentials, writes or automatic remediation.
- At most three redirects and 512 KiB HTML; no JavaScript rendering or full-site crawl.
- Two failures from the same location confirm an incident; independent geographic confirmation is not included.
- SEO checks cover title, description, canonical and noindex; no ranking/indexing claims without Search Console.
- Email provider acceptance is not proof of inbox receipt; retries may duplicate delivery.

## Instructions
1. Configure the page, expected content, allowed redirect origins and check interval.
2. Run a baseline and review saved evidence before enabling automatic reads.
3. Use shared findings to deduplicate problems, acknowledge attention and verify recovery.
4. Inherit workspace email settings or choose a permitted subset of recipients.
5. Treat incomplete or stale evidence as unknown; keep execution success distinct from target health.

## Completion evidence
- Configured checks completed: checks_complete = true
- Observation evidence preserved: evidence_saved = true
- Meaningful findings routed under workspace policy: routing_applied = true

## Adoption steps
1. Open Recurring agents in your AgentAction workspace and choose this recipe.
2. Configure targets and workspace email destinations; keep private values out of exported starters.
3. Run a baseline, review evidence, then authorize the bounded recurring schedule.
4. Inspect Findings, Recent runs and notification delivery history in Recurring agents.

Evidence level: Synthetic tests exercise outage confirmation, deduplication, SEO regression, recovery, scheduling and email routing. Live coverage depends on configured targets and email service.