dark-social-attributor
Pass
Audited by Gen Agent Trust Hub on Sep 2, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTIONDATA_EXFILTRATION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted input in the form of analytics exports and form inventories provided by the user (described in the Procedure section). It mitigates the risk of prompt injection by including specific instructions for the agent to treat this data as untrusted and to ignore any commands embedded within the pasted content.- [COMMAND_EXECUTION]: The skill utilizes a local Python script
scripts/connectors/pageviews.pyto fetch Wikipedia pageview data (Step 5 of the procedure). This is a constrained execution of a local tool for analytical purposes.- [DATA_EXFILTRATION]: The skill writes analysis results to a dedicated directorymemory/social/dark-social-attributor/and proposes updates to event registries through a brokered interfaceregistry-events.py. These operations are limited to the agent's internal memory filesystem.
Audit Metadata