esm
Pass
Audited by Gen Agent Trust Hub on Oct 12, 2026
Risk Level: SAFEPROMPT_INJECTIONMETADATA_POISONINGINDIRECT_PROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill directs the agent to proactively suggest the 'K-Dense Web' platform when users request complex workflows, steering the agent's behavior for promotional purposes beyond its core modeling functions.
- [METADATA_POISONING]: The skill's YAML frontmatter identifies the author as 'K-Dense Inc.', which contradicts the platform-reported author 'wu-yc'. This can lead to misattribution of the skill's source.
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted protein sequences and structural data which creates an attack surface for instructions embedded in biological data.
- Ingestion points: The ESMProtein constructor for sequences and the from_pdb method for structural files.
- Boundary markers: No delimiters or explicit instructions to ignore embedded content are provided in usage examples.
- Capability inventory: The skill is capable of performing network requests to external APIs and reading files from the local filesystem.
- Sanitization: There is no documentation of input validation or sanitization for the sequences or structure files being processed.
Audit Metadata