agent-rules
Pass
Audited by Gen Agent Trust Hub on Mar 27, 2026
Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [PROMPT_INJECTION]: Indirect Prompt Injection Surface. The skill enables the creation of persistent instructions that the agent loads automatically in future sessions via the modification of CLAUDE.md and AGENTS.md.
- Ingestion points: User-supplied imperatives (rules) enter the system through the Save operation defined in SKILL.md.
- Boundary markers: None. Rules are stored as raw Markdown sections (## Entry Title) without delimiters or instructions to prevent the agent from obeying embedded malicious commands.
- Capability inventory: The skill performs file system writes, creates directories, and modifies core agent configuration files (CLAUDE.md and AGENTS.md).
- Sanitization: No input validation, escaping, or sanitization is performed on the rule content before it is written to the file system.
- [COMMAND_EXECUTION]: Local Shell Command Execution. The skill uses shell commands to manage the rule directory structure during the category creation process.
- Evidence: Executes 'mkdir -p agent-docs/rules/' within the Save operation workflow.
Audit Metadata