pi-prompting
Pass
Audited by Gen Agent Trust Hub on Sep 14, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill consists entirely of instructional markdown content providing guidance for prompt engineering. It does not contain executable scripts, remote downloads, or network exfiltration patterns.
- [SAFE]: The documentation encourages the use of XML-style tag blocks (e.g.,
<task>,<grounding_rules>) as stable structural elements. These act as boundary markers that help distinguish instructions from processed data, reducing the likelihood of accidental obedience to instructions embedded within the data (Indirect Prompt Injection). - [SAFE]: The recommendation to include an
<action_safety>block for write-capable tasks, specifically naming allowed directories and file types, promotes the security principle of least privilege by explicitly restricting the scope of potential modifications.
Audit Metadata