Discovery setup

RSS and optional Serper search collect unverified source-document leads. Maintainers review them before incident intake. No automatic merge or eligibility promotion occurs.

Use RSS locally

python3 -I scripts/incidents/collect.py
python3 -I scripts/incidents/collect.py --execute \
  --source google-news-ethereum --source openzeppelin
python3 -I scripts/incidents/health.py --root incidents/discovery

The first command is an offline plan. The second explicitly makes bounded feed requests. Status records failures and limits; it cannot guarantee complete coverage.

Optional Serper

Provide SERPER_API_KEY through your environment or secret manager, never as a CLI argument. The collector does not read .env.

python3 -I scripts/incidents/collect.py --execute \
  --source serper-news --source serper-x-links --max-items 10 --days 7 \
  --status-out incidents/discovery/serper-status.json
python3 -I scripts/incidents/health.py --root incidents/discovery --include-serper

Serper searches news and indexed public X links. No direct X requests or browser scraping occurs. Copied post/article bodies are not retained. Unknown publication dates stay null; observation time is not earliest disclosure.

GitHub schedule

The prepared workflow checks every six hours and opens a draft review PR. It starts only after publication on the default branch and enabling Actions. Optional Serper uses repository secret SERPER_API_KEY and variable ENABLE_SERPER_DISCOVERY=true. Schedules can be delayed or disabled by GitHub.

See the complete operations guide for request limits, queue preservation, health, token behavior and failure handling.