DEV Community

SorrelThorncroft
SorrelThorncroft

Posted on

Patrowl's two app matches and 79 title matches need different review paths

Patrowl's two app matches and 79 title matches need different review paths

Patrowl is an open-source security operations orchestration platform. Its manager coordinates scanners and security workflows, while the engine API performs the individual checks. That separation can create different network paths and different authentication assumptions, so one external count cannot describe the whole deployment.

Observation

At 07:51 on November 20, 2026 (UTC+8), a count-only ZoomEye SDK query used sub_type=all, page 1, and pagesize 1. app="Patrowl" returned 2 matches. title="Patrowl" returned 79 matches.

The application fingerprint is the smaller and more specific signal. The title count is useful for discovery but may include documentation, marketing pages, and instances whose title is more generic. The two populations should not be merged into a single confirmed-server count.

Manager and engine trust

Patrowl's engine API documentation describes a generic JSON API used by the manager. Public descriptions of the engine note that it is commonly designed to run on localhost, and one API catalog states that the engine API has no authentication by default. That makes network placement important. If the engine listener is reachable from an untrusted network, the manager's authentication does not necessarily protect the engine directly.

The manager documentation and product descriptions describe RBAC, multiple users, teams, and multi-tenancy. Review who can create assets, start scans, view findings, and configure scanner engines. Also review the token used by the manager to call an engine and the engine's own exposure.

Build an asset-and-agent map

For each verified manager, list its teams, users, API tokens, scanner engines, target ranges, and result retention. For each engine, record its listener address, network policy, manager identity, and direct access rules. Confirm that a direct request to an engine cannot bypass the manager's authorization model.

Test with a user from one team attempting to read another team's scan, a token with no management role, an engine request from an unapproved host, and a stale scanner credential. Check whether failures are logged with a stable reason.

The app count of 2 is a high-confidence starting list for fingerprint validation. The title count of 79 is a broader ownership queue. Both are observations from one bounded query, not a statement that either host is vulnerable.

References

Top comments (0)