competitive-analysis
Pass
Audited by Gen Agent Trust Hub on Oct 11, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill identifies a vulnerability surface for indirect prompt injection, as it is designed to ingest and analyze untrusted data from user-supplied files and filings.
- Ingestion points: Untrusted data enters the agent context through user-provided source files used for extracting competitor metrics as specified in the 'Source files provided by the user' section of SKILL.md.
- Boundary markers: The instructions do not define explicit delimiters or use 'ignore embedded instructions' warnings to isolate external data from the agent's instructional context.
- Capability inventory: The skill leverages internal document and presentation generation tools (pptx and docx) to produce its deliverables, which represents a capability that could be influenced by malicious instructions in the source data.
- Sanitization: The workflow lacks sanitization or validation steps for content extracted from external sources before it is processed or included in output artifacts.
- Remediation: Implement structured data boundaries (e.g., XML tags) and explicit instructions to the agent to treat all content within those boundaries strictly as data and to ignore any natural language commands found therein.
Audit Metadata