oh-my-ppt-source-reading
Pass
Audited by Gen Agent Trust Hub on Aug 29, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to process external source documents and snippets, which is an inherent attack surface for indirect prompt injection. However, the skill implements several best-practice mitigations:
- Ingestion points: External data enters through
sourceDocumentPathsand retrieved snippets as defined inSKILL.md. - Boundary markers: The instructions explicitly command the agent to 'Do not discover, infer, or guess source document paths' and to 'Use only the sourceDocumentPaths explicitly provided by the host prompt'.
- Capability inventory: The skill uses targeted filesystem tools including
grep,glob, andread_fileto inspect sections of files. - Sanitization: The instructions mandate using 'literal string' patterns for
grepand progressively reading small sections (50-80 lines) rather than entire files, limiting the risk of malicious payload execution. - [SAFE]: The skill instructions focus on factual grounding and explicitly forbid inventing facts, metrics, or quotes. The implementation of 'Long-document discipline' and strict adherence to host-provided paths represent a secure approach to file-based workflows.
Audit Metadata