event-driven-outreach
Pass
Audited by Gen Agent Trust Hub on Oct 10, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill includes a Python utility script
scripts/check-output.pyintended for local execution by the agent. - Evidence: The script uses
sys.argv[1]to accept a file path and then callspath.read_text()to read the file's content. While intended for validating generated deliverables, this capability could be misused to read sensitive local files if the agent is not properly constrained. - [INDIRECT_PROMPT_INJECTION]: The skill's primary workflow involves ingesting and processing data from external, third-party sources.
- Ingestion points:
SKILL.mdinstructs the agent to source data from "attendee lists, speaker lists, sponsor lists, LinkedIn event RSVPs, hashtag engagers". - Boundary markers: The skill does not provide specific delimiters or instructions to treat the ingested event data as untrusted or to ignore embedded instructions.
- Capability inventory: The agent is expected to use web search/scraping tools to collect data and the provided
scripts/check-output.pyfor validation. - Sanitization: There is no evidence of sanitization, filtering, or validation of the content retrieved from external attendee lists or social media feeds.
Audit Metadata