claude-code-mastery

Pass

Audited by Gen Agent Trust Hub on Nov 8, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill ingests local filesystem data (such as CLAUDE.md content and directory structures) to provide analysis and optimization reports. It also accepts user input for skill metadata during scaffolding.
  • Ingestion points: claudemd_optimizer.py (reads project CLAUDE.md) and skill_scaffolder.py (CLI arguments).
  • Boundary markers: Standard template interpolation and file reading; no specific "ignore instructions" markers are used for processed data.
  • Capability inventory: skill_scaffolder.py creates new files using Path.write_text and sets executable permissions using os.chmod.
  • Sanitization: Basic validation of skill names is implemented in validate_skill_name to restrict characters to alphanumeric and hyphens.
  • [COMMAND_EXECUTION]: The documentation (references/hooks-cookbook.md) provides numerous shell command examples for automating tasks like formatting and linting. These serve as educational templates for users to configure their own developer environment and include safety-oriented recipes designed to block dangerous bash commands.
Audit Metadata
Risk Level
SAFE
Analyzed
Nov 8, 2026, 03:14 PM