`pnpm research:external-impact` imports row-level public evidence of a citation,
substantive reuse, or externally accepted correction. Aggregate traffic,
downloads, clicks, replies, AI activity, internal links, and operator assertions
do not count.
Every event must identify the exact reviewed investigation ID, version, and
packet SHA-256. The importer derives the investigation's review time from the
signed validation artifact, and every event must occur strictly after that
review and no later than evaluation time. Pre-release and future-dated events
fail closed. Each event also requires a unique external HTTPS URL, pseudonymous
external-actor code, observation time, and a separate substantive verification
receipt. Meta Museum and Sun and Rain Works domains are rejected, as are common
example, test, sandbox, demo, and fixture hosts. Localhost, private/link-local
IPv4, and private/loopback IPv6 evidence URLs are also rejected. Direct
identifiers are rejected recursively.
Run
`pnpm research:external-impact -- --validation-artifact=<validation-readiness.json> --events=<external-events.json>`.
The importer verifies the signed validation artifact, derives the exact
reviewed investigation ID/version/packet digest, and computes the event-file
receipt. It also derives the exact reviewed timestamp rather than trusting the
event file to declare when the release existed. Use `--output=<directory>` for a controlled run; configuration mode
remains available for scheduling. The resulting artifact is projected into A+
evidence only when its integrity digest verifies, and the A+ scorer repeats the
release cross-check. The default configuration is intentionally empty and
blocked.
This command performs no outreach, publication, recruitment, or external
mutation and cannot claim scholarly impact without real attributable evidence.
The signed artifact retains the complete reviewed-investigation binding and every
normalized citation, reuse, or accepted-correction event with its independent
verification receipt. Ingestion reruns the evaluator at the artifact's original
evaluation time and requires canonical equality. Editing a projected event count
or source and recalculating the outer hash cannot produce accepted impact evidence.