ChatGPT
Pass
Audited by Gen Agent Trust Hub on Sep 7, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill provides templates for processing user-provided context and source materials to construct ChatGPT prompts, creating a potential surface for indirect prompt injection.\n
- Ingestion points: External data is incorporated through context, source material, and project files as described in
SKILL.mdandprompt-packets.md.\n - Boundary markers: The skill explicitly promotes the use of structured 'Prompt Packets' in
prompt-packets.mdwith defined fields like Goal, Context, and Source of Truth to separate instructions from data.\n - Capability inventory: The skill's primary capability involves local file management within the
~/chatgpt/directory for storing project logs and workflow defaults (SKILL.md,memory-template.md).\n - Sanitization: Robust output verification is guided by
output-qa.md, which includes specific checks for source fidelity, constraint adherence, and self-critique prompts to validate responses before use.
Audit Metadata