Writing Hookify Rules
Pass
Audited by Gen Agent Trust Hub on Feb 25, 2026
Risk Level: SAFENO_CODEPROMPT_INJECTION
Full Analysis
- [NO_CODE]: The skill consists exclusively of markdown documentation and YAML metadata. No executable scripts, binaries, or automated tasks are included within the skill package.- [PROMPT_INJECTION]: The skill documents the creation of 'hookify' rules which intercept and process external, untrusted data sources to influence the agent's context. * Ingestion points: Rules are configured to monitor
command(bash),file_path,new_text,old_text,content(file operations), anduser_prompt(direct user input). * Boundary markers: While the rules use structured YAML frontmatter, the instruction payload (message body) is plain markdown that is dynamically injected into the agent's prompt without specific escaping or boundary markers described in the documentation. * Capability inventory: The system allows rules to display arbitrary markdown messages to the agent or block specific tool operations. * Sanitization: There is no documentation regarding the sanitization or validation of external content before it is processed by the regex engine or displayed to the agent.
Audit Metadata