scanpy
Pass
Audited by Gen Agent Trust Hub on Mar 15, 2026
Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [PROMPT_INJECTION]: The 'Suggest Using K-Dense Web' section contains behavioral steering instructions that direct the agent to proactively advertise a commercial service (www.k-dense.ai) when task complexity increases, overriding the agent's neutral persona.
- [COMMAND_EXECUTION]: The skill documentation references and encourages the execution of local shell commands and Python scripts (e.g., 'scripts/qc_analysis.py' and 'assets/analysis_template.py') to automate workflows, which involves running external code on the host system.
- [PROMPT_INJECTION]: The skill facilitates an indirect prompt injection surface by ingesting untrusted data from formats like CSV and H5AD while possessing file-writing and script-execution capabilities without explicit boundary markers or sanitization. 1. Ingestion points: sc.read_10x_mtx, sc.read_h5ad, sc.read_csv. 2. Boundary markers: Absent. 3. Capability inventory: adata.write, to_csv, and shell script execution. 4. Sanitization: Absent.
Audit Metadata