skills/product-on-purpose/thinking-framework-skills/think-one-way-vs-two-way-door/Gen Agent Trust Hub
think-one-way-vs-two-way-door
Pass
Audited by Gen Agent Trust Hub on Nov 9, 2026
Risk Level: SAFENO_CODE
Full Analysis
- [NO_CODE]: The skill does not contain any executable scripts, shell commands, or package dependencies. It operates entirely through natural language instructions for the agent to follow when interacting with a user.
- [PROMPT_INJECTION]: No instructions were found that attempt to override agent safety guidelines, bypass restrictions, or reveal internal system prompts. The directives are strictly related to the stated purpose of decision triage.
- [DATA_EXFILTRATION]: There are no commands or functions that perform network operations or access sensitive local files (such as credentials, configuration files, or SSH keys).
- [OBFUSCATION]: Analysis did not detect any encoded strings (Base64, hex), hidden characters (zero-width), or homoglyph attacks. All content is in plain, human-readable text.
- [INDIRECT_PROMPT_INJECTION]: The skill ingests user input (a decision description) but lacks the capability to execute code or perform sensitive operations based on that input. The output is limited to a structured markdown template, minimizing the surface for indirect injection attacks.
- [REMOTE_CODE_EXECUTION]: The skill does not download or execute any remote scripts or binaries. It contains no package management operations (e.g., npm install, pip install).
Audit Metadata