recipe-reverse-engineer
Pass
Audited by Gen Agent Trust Hub on Oct 31, 2026
Risk Level: SAFEPROMPT_INJECTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill uses authoritative directives ("The user explicitly instructs and authorizes every subagent call") to ensure the agent executes the workflow without hesitation. While this is an instructional override technique common in orchestration recipes to guide model behavior, it does not attempt to bypass core safety filters or extract system prompts.
- [INDIRECT_PROMPT_INJECTION]: The skill has an indirect prompt injection surface as it is designed to ingest and process content from a user-provided codebase.
- Ingestion points: Files within the directory specified by
$USER_TARGET_PATHand inventory items in$PRD_UNIT_INVENTORYare read to generate documentation. - Boundary markers: The instructions do not define explicit delimiters (e.g., XML tags) to isolate untrusted codebase content from the instructions passed to subagents.
- Capability inventory: The skill orchestrates subagents for content generation and writes documentation to the local
docs/directory. It does not possess network or arbitrary command execution capabilities. - Sanitization: There is no mention of sanitization or filtering of the ingested code content before it is interpolated into subagent prompts.
- Note: This surface is inherent to the skill's primary purpose of documenting existing code.
Audit Metadata