military-manual
Pass
Audited by Gen Agent Trust Hub on Mar 14, 2026
Risk Level: SAFE
Full Analysis
- [PROMPT_INJECTION]: The skill utilizes a military persona and authoritative language (e.g., "Strictly execute", "Any slack must be corrected immediately"). This content is consistent with the stated theme of a 'Military Manual' for software development and does not target the underlying safety or ethical guidelines of the AI.
- [DATA_EXPOSURE]: No hardcoded credentials, API keys, or access to sensitive system file paths (such as SSH keys or environment variables) were detected within the instructions or metadata.
- [EXTERNAL_DOWNLOADS]: The skill references internal markdown files (e.g., '军事化组织·侦察兵情报收集.md') as context for its roles. These are local references and do not involve downloading code from untrusted remote servers.
- [COMMAND_EXECUTION]: There are no scripts, shell commands, or subprocess calls included in the skill. All instructions are purely textual and intended for the AI's natural language processing.
- [INDIRECT_PROMPT_INJECTION]: The skill defines an input surface for '作战目标' (Project Objectives). While this allows the ingestion of untrusted user data, the skill lacks high-privilege capabilities (like network access or file system modification) that would make such an injection exploitable. This is a standard functional surface with minimal risk.
Audit Metadata