prompting-gpt
Pass
Audited by Gen Agent Trust Hub on Sep 1, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill serves as a documentation and guidance tool for prompt engineering, emphasizing structure, constraints, and validation.
- [SAFE]: References to external repositories (e.g., github.com/openai/skills) and APIs (api.openai.com) target trusted organizations and well-known services.
- [SAFE]: No obfuscation, persistence mechanisms, or unauthorized privilege escalation patterns were identified.
- [INDIRECT_PROMPT_INJECTION]: The skill acknowledges the risk of processing untrusted data and proactively instructs the agent to define safe autonomous actions, approval boundaries, and evidence requirements to mitigate potential injection or misbehavior. Ingestion points: SKILL.md (prompts, tool descriptions), gpt-5.6.md (user inputs). Boundary markers: Present (e.g., <output_contract>, <tool_persistence_rules>). Capability inventory: gpt-5.4.md (code_exec, computer use), gpt-5.6.md (programmatic_tool_calling, terminal tool). Sanitization: Recommended via Structured Outputs.
Audit Metadata