TaskSignal is for research, product discovery, and learning.
Principles:
- Use public data only.
- Prefer official APIs over scraping.
- Do not store raw usernames by default.
- Store
author_hashornull. - Preserve source URLs for attribution.
- Respect rate limits.
- Avoid spam, harassment, or manipulation workflows.
- Collect only what is needed for opportunity research.
Live scans use official connector APIs where available. The scan pipeline stores
normalized author_hash values instead of raw usernames and keeps source URLs so
ranked opportunities remain attributable. Connector responses are minimized
before storage so raw author fields from live APIs are not retained in
raw_items.
Set AUTHOR_HASH_SALT to a machine-specific value before storing live scan
results. The readiness endpoint warns when the placeholder salt is still in use
without returning the configured value.
Review source limits and terms before enabling a connector
in a hosted deployment. The public scan API only exposes non-credentialed
sources and can be narrowed further with PUBLIC_SCAN_SOURCES; credentialed
sources belong in trusted internal jobs after review.
TaskSignal does not include outreach automation. Use scan results for research, product discovery, and evidence review, not bulk replies or targeting people.
For local demos:
make reset-dataFor hosted deployments, add administrative deletion endpoints or database retention policies before collecting live data.