bio-binning-qc
Pass
Audited by Gen Agent Trust Hub on Oct 5, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADSCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill involves a data processing surface where it ingests outputs from various bioinformatic tools to generate summary tables.
- Ingestion points: The
scripts/build_bin_qc_tables.pyscript reads multiple TSV files (CheckM2, GUNC, EukCC, GTDB-Tk) provided as command-line arguments. - Boundary markers: None; data is processed as structured TSV/CSV format.
- Capability inventory: The skill performs filesystem writes to a
results/directory via the Python script and tool executions. - Sanitization: The Python script uses the standard
csv.DictReaderfor parsing and includes logic to validate domain routes and ensure identifiers are present, reducing the risk of processing malformed or malicious data structures. - [EXTERNAL_DOWNLOADS]: The skill provides instructions for fetching necessary scientific databases and software environments from reputable sources.
- Downloads the CheckM2 database from Zenodo (DOI 10.5281/zenodo.14897628).
- Fetches the EukCC2 database from the European Bioinformatics Institute (EBI) official FTP servers.
- Pulls the BBTools container from the official repository maintained by Bryce Foster (
bryce911/bbtools). - [COMMAND_EXECUTION]: The skill documentation includes examples of executing bioinformatic tools via Docker and local shell commands.
- The recommended Docker command (
docker run --rm -u "$(id -u):$(id -g)" -v "$PWD":/work ...) utilizes non-root user mapping and restricted volume mounting, which are standard security best practices for containerized workflows to prevent privilege escalation or unintended filesystem modifications.
Audit Metadata