deep-sky-processor

Pass

Audited by Gen Agent Trust Hub on Sep 23, 2026

Risk Level: SAFECOMMAND_EXECUTIONPRIVILEGE_ESCALATIONINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [COMMAND_EXECUTION]: The scripts/star_tools.py and scripts/plate_solve.py components execute external command-line tools (StarNet2 and Astrometry.net) using subprocess.run. The paths to these binaries are provided via command-line arguments or environment variables, allowing for the execution of arbitrary local files if the agent is misled.
  • [PRIVILEGE_ESCALATION]: scripts/star_tools.py includes logic to modify file permissions. It calls chmod +x on the user-provided path for the StarNet2 binary to ensure it can be run. Additionally, on macOS, it attempts to remove the com.apple.quarantine attribute from the binary's parent directory using xattr, which bypasses default operating system security protections for third-party software.
  • [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted metadata from astronomical data files (FITS/XISF).
  • Ingestion points: Metadata fields such as OBJECT and TARGNAME are read from file headers in scripts/fits_io.py and used to drive processing logic.
  • Boundary markers: There are no explicit delimiters used to separate this metadata from the agent's core instructions.
  • Capability inventory: The skill has significant capabilities, including local shell execution via subprocess.run and network access.
  • Sanitization: No sanitization is performed on the ingested header strings before they are processed or logged.
  • [EXTERNAL_DOWNLOADS]: scripts/fits_io.py performs network requests to the CDS Sesame astronomical service at http://cdsweb.u-strasbg.fr to resolve celestial target names into coordinates. This is a standard lookup for astronomical data but involves communicating with an external domain.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 23, 2026, 12:17 PM