rsvp-comprehension-agent

Warn

Audited by Gen Agent Trust Hub on Apr 3, 2026

Risk Level: MEDIUMDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
  • [DATA_EXFILTRATION]: The file hooks contains a directory traversal string (../../hooks). This pattern indicates an attempt to reference or access directories outside the skill's defined scope, which could potentially expose sensitive system information or configuration files located in parent directories.
  • [PROMPT_INJECTION]: The skill presents an indirect prompt injection surface due to its handling of external data for session management and quiz generation.
  • Ingestion points: The skill reads JSON token streams using the Read tool as defined in SKILL.md.
  • Boundary markers: There are no instructions or delimiters specified to isolate the untrusted token stream content from the agent's instructions, nor are there warnings to ignore embedded commands.
  • Capability inventory: The agent is granted access to the Bash, Read, and Write tools, which provides a significant capability tier that could be exploited via malicious content in the processed data.
  • Sanitization: The skill instructions do not specify any validation, filtering, or escaping of the token stream content before it is processed by the model for comprehension checks.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Apr 3, 2026, 06:09 PM