product-strategy-session

Pass

Audited by Gen Agent Trust Hub on Sep 14, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process external data, such as context dumps or existing discovery artifacts, which could contain malicious instructions designed to alter the agent's behavior.\n
  • Ingestion points: Data enters the skill context through user-supplied text at invocation, pasted 'context dumps', and content appended via an ARGUMENTS: line in SKILL.md.\n
  • Boundary markers: The skill lacks explicit delimiters or instructions to the agent to treat user-provided strategic artifacts as untrusted data, increasing the risk that embedded commands could be followed.\n
  • Capability inventory: While SKILL.md does not directly invoke dangerous system commands, it orchestrates a complex workflow involving 15+ sub-skills. The combined capability of the orchestrated environment could be influenced by injected data.\n
  • Sanitization: No input validation, escaping, or filtering mechanisms are implemented for the external product artifacts processed by the workflow.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 14, 2026, 02:36 PM
Security Audit — agent-trust-hub — product-strategy-session