Agency Evidence Pack

For recurring client reporting, Agency Evidence Pack builds a multi-client observed snapshot and change report from existing Checksmith public-site signal engines.

It is designed for agencies or operators who need a repeatable monthly record across multiple client sites without manually stitching reports together.

The Actor accepts one to twenty public client sites. It runs the same bounded checks for each client, isolates failures to the affected engine or client, and returns one consistent evidence pack. A failed observation does not become a positive finding and one unreachable site does not prevent the remaining clients from completing.

What It Produces

• Client-by-client observed snapshot
• Engine status per client
• Previous-snapshot comparison when provided
• Client-level Evidence Pack references
• Markdown, HTML, JSON, and dataset rows
• Fixed boundary disclaimer in the report footer

The current engine set records public AI-crawler readiness signals and public domain, TLS, DMARC, MTA-STS, and security.txt deadline signals. Each engine reports a checked, unobserved, or error state. The pack does not combine those observations into an agency score or a client quality grade.

Typical Workflow

1. Add each client name, bare domain, and public site URL.
2. Run the Actor before a recurring review or client handoff.
3. Download REPORT.html, REPORT.md, or REPORT.json from the run storage.
4. Keep currentSnapshot if the next run should label clients as new, changed, or unchanged.

The previous snapshot is supplied explicitly. The Actor does not silently maintain a long-term client database and does not require an account login for any client site.

Billing

• agency-evidence-pack-generated: USD 3.00 once when at least one client has an observable result.
• agency-client-checked: USD 1.20 for each client with at least one observable engine result.

A one-client run is USD 4.20, five checked clients are USD 9.00, and the maximum twenty-client run is USD 27.00, plus Apify platform usage. Invalid input and runs with no observable client result do not trigger either custom event. A client for which every engine errors or returns only unobserved signals is not counted as checked.

Boundaries

This product does not provide SEO, security, compliance, uptime, agency-performance, ranking, traffic, or platform-acceptance certification. It does not prove that a site is managed correctly. It reports machine-observed public-site signals and changes only.

Template editing is intentionally not supported. Branding is limited to agency name, optional logo URL metadata, and an optional footer note while the boundary disclaimer remains included.

The report is not a replacement for manual SEO, security, legal, compliance, uptime, traffic, or platform review. It does not log into client systems, repair findings, monitor private networks, predict rankings, or certify that an agency has managed a site correctly. The logo URL is stored as metadata only; it is not fetched or embedded by the Actor.

Support Boundary

Supported questions cover input shape, public URL requirements, report locations, snapshot handoff, and reproducible Actor errors. Support does not include interpreting whether a client site is good, providing remediation services, editing client reports, promising platform outcomes, or reviewing private account data.

Local Example

$PYTHONPATH=src python -m agency_evidence_pack.cli examples/sample-input.json --json report.json --md report.md

For a first run, keep sampleLimit small and use public test domains. Review the generated engine statuses before adding more clients.