aif-rules
Pass
Audited by Gen Agent Trust Hub on Sep 18, 2026
Risk Level: SAFEPROMPT_INJECTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The instructions explicitly mandate that the agent prioritize external content from '.ai-factory/skill-context/aif-rules/SKILL.md' over its own system instructions. It uses override language such as 'the skill-context rule wins' and 'Do NOT ignore skill-context rules even if they seem to contradict this skill's defaults', which functions as a built-in instruction bypass mechanism.
- [INDIRECT_PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection due to its requirement to load and obey instructions from external project files without sanitization. -- Ingestion points: Skill context file at '.ai-factory/skill-context/aif-rules/SKILL.md' and user-supplied rule text. -- Boundary markers: Absent; content from external files is granted higher priority than the skill's own rules. -- Capability inventory: The skill possesses 'Write' and 'Edit' capabilities to modify project configuration and documentation artifacts. -- Sanitization: No validation or sanitization of ingested instructions is implemented.
Audit Metadata