The Agent Skills Directory

[PROMPT_INJECTION]: No instructions were detected that attempt to override agent safety protocols, bypass guidelines, or extract system prompts. The instructions focus on logical classification and critical questioning.- [DATA_EXFILTRATION]: No hardcoded credentials, sensitive file paths, or network exfiltration patterns were identified. A visible URL to the author's GitHub repository is present but involves no exfiltration logic.- [OBFUSCATION]: The skill uses clear, plain-text markdown and YAML. No base64 encoding, zero-width characters, homoglyphs, or other obfuscation techniques were found.- [REMOTE_CODE_EXECUTION]: No remote scripts or unverifiable dependencies are downloaded or executed.- [DYNAMIC_EXECUTION]: No patterns involving dynamic code generation, eval(), or unsafe deserialization were detected.- [INDIRECT_PROMPT_INJECTION]: The skill ingests user-supplied arguments for analysis but has no dangerous capabilities (such as network access or file system modification) that could be triggered by malicious data. Mandatory evidence chain: 1. Ingestion point: Argument to be evaluated (SKILL.md); 2. Boundary markers: Instruction to restate the argument (SKILL.md); 3. Capability inventory: The skill uses no external tools or scripts; 4. Sanitization: Not applicable as no code is executed.- [DYNAMIC_CONTEXT_INJECTION]: The skill does not use the shell command execution syntax (!command) in its definition files.

think-walton-argumentation-schemes