The Agent Skills Directory

[SAFE]: The skill facilitates the processing of Flow Cytometry Standard (FCS) files using established Python libraries such as flowio, numpy, and pandas. All code examples demonstrate local file operations and standard data analysis workflows. No patterns of credential theft, unauthorized network exfiltration, obfuscation, or malicious command execution were identified.
[INDIRECT_PROMPT_INJECTION]: The skill parses external FCS files which include metadata keywords (e.g., instrument names, acquisition dates). While this represents a potential surface for indirect instructions if an agent processes this metadata as natural language, the risk is inherent to file parsing and no specific vulnerabilities were identified.
Ingestion points: FlowData constructor (in SKILL.md and api_reference.md) reads content from external FCS files.
Boundary markers: None provided in the prompt templates.
Capability inventory: File system write capabilities via create_fcs() and write_fcs() methods documented in SKILL.md and api_reference.md.
Sanitization: None explicitly mentioned for the metadata fields extracted from the TEXT segment of FCS files.

flowio