alterlab-omero
Pass
Audited by Gen Agent Trust Hub on Sep 25, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill retrieves metadata, annotations, and tabular data from external OMERO servers. This content is considered untrusted and is introduced into the agent's context without sanitization or explicit boundary markers. A malicious OMERO server or an attacker with write access to the server could theoretically attempt to influence the agent's behavior through crafted metadata. This risk is notable given the powerful capabilities granted to the skill in its platform configuration.\n
- Ingestion points: Multiple retrieval methods across
references/data_access.md,references/metadata.md, andreferences/tables.md(e.g.,image.getName(),map_ann.getValue(),table.read()).\n - Boundary markers: None identified in the documentation or provided code samples for handling server-retrieved strings.\n
- Capability inventory: The skill has access to
Bash(uv:*),Write, andEdittools as defined inSKILL.md.\n - Sanitization: There is no description or implementation of sanitization or escaping for metadata strings retrieved from the OMERO server.
Audit Metadata