Panda Suite Docs
v1.9.2GitHub ↗
Panda Suite / Blue scheduling

Scheduled intelligence refresh (optional)

Nothing is scheduled by default. Panda Blue does not install a scheduler, run a daemon, send alerts or claim continuous monitoring. Use this page only if you want the intelligence cache refreshed without asking an agent each time. Every command here is an illustrative setup for you to review and install yourself.

What a scheduled run does

intel update contacts the fixed public endpoints in the source registry: NVD, CISA KEV, the three OWASP pages and the MITRE ATT&CK index. It writes dated snapshots to the private cache. It sends no source code, paths or tokens. Package names and versions go to OSV only when you add --allow-package-queries and a component inventory.

Each run fetches every selected source. It does not skip sources whose cache is still fresh. To check OWASP and ATT&CK only weekly, schedule them separately with --source.

Missed runs catch up: NVD resumes from the last successful sync (minus two minutes). If the gap holds more than 8,000 changes it is refused; see the NVD note under Known limits for the one-command restart. A failed source keeps its previous snapshot, and the run exits 3.

Local scheduler (cron)

Run from the Suite checkout so the default cache .panda-suite/blue/intelligence-cache is used. Example crontab entries:

# Daily at 06:00: KEV and NVD changes
0 6 * * * cd /path/to/panda-suite && .venv/bin/python skills/panda-blue/scripts/blue.py intel update --source kev --source nvd --allow-network >> .panda-suite/blue/intel-update.log 2>&1
# Mondays at 06:30: OWASP and ATT&CK versions
30 6 * * 1 cd /path/to/panda-suite && .venv/bin/python skills/panda-blue/scripts/blue.py intel update --source owasp-web --source owasp-api --source owasp-asvs --source mitre --allow-network >> .panda-suite/blue/intel-update.log 2>&1

On macOS, launchd works the same way with the same command. Check results at any time without network access:

.venv/bin/python skills/panda-blue/scripts/blue.py intel status --offline

Exit 0 means every enabled source is fresh. Exit 3 means at least one is stale, unavailable or failed; sourceStatus says which and why.

Project exposure summary

To also check a project's npm dependencies, add the inventory and the separate package permission. Reports need a new directory each run:

0 7 * * * cd /path/to/panda-suite && .venv/bin/python skills/panda-blue/scripts/blue.py intel latest --allow-network --allow-package-queries --project /path/to/project --output .panda-suite/blue/reports/intel-$(date +\%Y\%m\%d) >> .panda-suite/blue/intel-update.log 2>&1

Reports stay in the private .panda-suite/ directory. Nothing is published, emailed or used to upgrade dependencies. Review them before sharing.

GitHub Actions

A scheduled workflow can run the same commands, but consider privacy first: a workflow on a public repository exposes its logs, and caches or artifacts can reveal your dependency list. If you use one, keep the repository private, grant permissions: contents: read only, do not upload reports as public artifacts, and do not let the workflow open pull requests or upgrade packages automatically. Panda Suite ships no such workflow.

Known limits

  • NVD change volume: about 1,400 NVD records change per day. One run reads at most 8,000 (8 pages of 1,000, 6 seconds apart). A daily update stays well inside that. After roughly five or more missed days the catch-up is refused with NVD_WINDOW_TOO_LARGE, and the previous snapshot is kept. To recover, run intel update --source nvd --days 1 --allow-network once. That restarts the sync one day back, and the output says that older changes were not fetched. Later default runs resume from there. Nothing is silently truncated.
  • Duplicate alert suppression is not provided because no alerts are sent.
  • Unauthenticated NVD requests are rate limited; the helper retries with backoff and respects Retry-After, but frequent schedules make failures more likely.