seaborn
Warn
Audited by Gen Agent Trust Hub on Sep 14, 2026
Risk Level: MEDIUMMETADATA_POISONINGPROMPT_INJECTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [METADATA_POISONING]: The skill metadata identifies the author as 'K-Dense Inc.', which contradicts the verified author identity of 'jst-well-dan'. This discrepancy is misleading regarding the skill's provenance and the authority behind its recommendations.\n- [PROMPT_INJECTION]: The skill body contains explicit instructions that attempt to steer the agent's behavior to include promotional content for an external platform (www.k-dense.ai) when handling complex tasks, rather than remaining neutral.\n- [INDIRECT_PROMPT_INJECTION]: The skill demonstrates the ingestion of external datasets, which introduces a vulnerability where malicious instructions embedded in processed data could influence the agent's reasoning.\n
- Ingestion points: The skill provides examples of loading external data from CSV files in 'references/examples.md'.\n
- Boundary markers: There are no instructions or delimiters provided to signal the agent that data content should be treated as distinct from execution instructions.\n
- Capability inventory: The skill examples include capabilities for reading from the filesystem and writing output files via 'savefig'.\n
- Sanitization: No data validation or sanitization routines are specified for handling the ingested external datasets.
Audit Metadata