The Agent Skills Directory

[SAFE]: The skill is a legitimate administrative extension for PostHog. It correctly identifies the scope of its audit and routes out-of-scope issues to the appropriate handlers. All tools referenced (e.g., view-list, view-run-history) are consistent with the vendor's existing infrastructure.
[INDIRECT_PROMPT_INJECTION]: The skill ingests external data such as materialized view names and error logs from the PostHog API. While this presents a theoretical injection surface, the risk is mitigated by explicit instructions to the agent to never perform destructive actions (like view-unmaterialize or view-delete) or modify queries without explicit human confirmation. The processing of this data is central to the skill's primary purpose.

auditing-warehouse-view-health