Testing Locally

You can test your scanner directly without installing it into HiveScanner.

Test the Configure Command

echo '{"command": "configure"}' | python community/your-scanner/adapter.py --sandboxed

Expected output — valid JSON with your default config:

{
  "config": {
    "enabled": false,
    "token_env": "YOUR_TOKEN",
    "max_items": 20
  }
}

Test Polling

echo '{"command": "poll", "config": {"enabled": true, "your_option": "value"}, "watermark": "1970-01-01T00:00:00Z"}' \
  | python community/your-scanner/adapter.py --sandboxed

Use 1970-01-01T00:00:00Z only to exercise initial state. A correct scanner normally bootstraps quietly, so the first poll should snapshot current provider state and return no historical flood.

Expected output — valid JSON with pollen items:

{
  "pollen": [
    {
      "id": "scanner-item-123",
      "source": "your-scanner",
      "type": "new_item",
      "title": "Something happened",
      "preview": "Details...",
      "discovered_at": "2025-01-15T10:30:00Z",
      "author": "user",
      "author_name": "User Name",
      "group": "Group",
      "url": "https://example.com",
      "metadata": {}
    }
  ],
  "watermark": "2025-01-15T10:30:00Z"
}

Checklist

Before submitting your scanner, verify:

• [ ] configure command returns valid JSON
• [ ] poll command returns valid JSON with all required pollen fields
• [ ] Pollen IDs are unique and use a scanner-specific prefix
• [ ] Watermarks advance correctly (no duplicate items on subsequent polls)
• [ ] First poll is quiet; configuration/identity changes do not replay old data
• [ ] Equal-timestamp delayed items are not skipped
• [ ] Duplicate IDs, repeated cursors, reordered pages, provider errors, and malformed late records fail closed
• [ ] Pagination/rate/time budgets preserve the old watermark instead of skipping backlog
• [ ] Provider text that looks like instructions remains inert data
• [ ] Errors are handled gracefully — return ([], watermark) on failure
• [ ] No secrets are hardcoded — all tokens use env var references
• [ ] teammate.json is valid and all fields are populated
• [ ] Only Python stdlib is used (no pip dependencies)