code-signing

Pass

Audited by Gen Agent Trust Hub on Jul 2, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSCREDENTIALS_UNSAFEPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill includes shell scripts (sign-macos.sh, sign-windows.sh) that execute system utilities such as osslsigncode, codesign, xattr, and PlistBuddy. These commands are used to modify file attributes and embed digital signatures into binaries.
  • [CREDENTIALS_UNSAFE]: The sign-windows.sh script and SKILL.md instructions contain a hardcoded default password (hyperspace2024) for the PKCS#12 certificate bundle. While part of a documented self-signing workflow, hardcoding passwords in scripts is an insecure practice.
  • [EXTERNAL_DOWNLOADS]: The skill references several well-known timestamping services (DigiCert, Sectigo, GlobalSign, Symantec) to ensure the longevity of digital signatures.
  • [PROMPT_INJECTION]: The helper scripts provide a surface for indirect prompt injection by processing external data (file paths and application names).
  • Ingestion points: CLI arguments $1, $2, and $3 in sign-macos.sh and sign-windows.sh.
  • Boundary markers: None present in the scripts or instructions.
  • Capability inventory: Subprocess calls to osslsigncode, codesign, xattr, PlistBuddy, and mv.
  • Sanitization: Variables are wrapped in double quotes in the shell commands, providing basic protection against some shell injection patterns but lacking comprehensive validation of the input content.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 2, 2026, 07:11 PM
Security Audit — agent-trust-hub — code-signing