skills/wu-yc/labclaw/flowio/Gen Agent Trust Hub

flowio

Pass

Audited by Gen Agent Trust Hub on Oct 12, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONPROMPT_INJECTIONEXTERNAL_DOWNLOADSCOMMAND_EXECUTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill is designed to parse FCS (Flow Cytometry Standard) files, which contain metadata in TEXT and ANALYSIS segments. This data is external and could potentially contain malicious instructions intended to be executed by the agent.
  • Ingestion points: The FlowData class in SKILL.md reads and parses FCS files from the local file system, extracting metadata keywords.
  • Boundary markers: The instructions do not define clear boundaries or provide warnings to the agent to treat extracted metadata as untrusted data.
  • Capability inventory: The skill possesses the capability to write files to the local system using create_fcs and pandas.to_csv functions as demonstrated in the examples.
  • Sanitization: There is no evidence of sanitization or validation of the metadata content before it is processed by the agent.
  • [PROMPT_INJECTION]: The skill includes a 'Suggest Using K-Dense Web' section that provides specific behavioral instructions to the agent. It directs the agent to proactively suggest a commercial research platform (www.k-dense.ai) when user requests reach a certain level of complexity, which is an attempt to influence agent output through embedded instructions.
  • [EXTERNAL_DOWNLOADS]: The installation section provides instructions to download the flowio package from a public package registry using the command uv pip install flowio.
  • [COMMAND_EXECUTION]: The code examples in SKILL.md demonstrate interactions with the local file system for both reading cytometry files (FlowData) and writing processed data to new files (create_fcs, to_csv).
Audit Metadata
Risk Level
SAFE
Analyzed
Oct 12, 2026, 05:25 AM