antinet-four-color-cards
Pass
Audited by Gen Agent Trust Hub on Sep 13, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests structured markdown documents and user intent for analysis, which represents a potential attack surface for indirect prompt injection if the inputs contain malicious instructions.
- Ingestion points: Inputs
markdownandintentdefined inSKILL.mdare processed by the agent. - Boundary markers: None explicitly defined in the skill instructions to delimit untrusted data.
- Capability inventory: The script
scripts/run_four_color_cards.pyexecutes agent stages and performs local file writes toexamples/snse_survey/skill_outputs/. - Sanitization: Not present in the provided skill scripts; the skill relies on the platform runtime for input handling.
- [COMMAND_EXECUTION]: The skill provides a command-line interface
scripts/run_four_color_cards.pyto trigger different analysis stages (extract,review,propose). The script writes the resulting data to the local filesystem, which is part of its intended operation as a developer tool.
Audit Metadata