git-guardrails-claude-code
Pass
Audited by Gen Agent Trust Hub on Aug 27, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill requires the agent to modify local application configuration files (
.claude/settings.jsonor project-specific.claude/settings.json) to register a new pre-execution hook.\n- [COMMAND_EXECUTION]: The installation procedure involves setting execution permissions (chmod +x) on the bundledscripts/block-dangerous-git.shscript to enable its functionality as a system hook.\n- [INDIRECT_PROMPT_INJECTION]: The skill implements a security filter that processes tool execution commands to detect and intercept potentially destructive operations.\n - Ingestion points: The hook script (
scripts/block-dangerous-git.sh) ingests input from standard input containing the agent's intended tool command.\n - Boundary markers: The script expects input in a structured JSON format and utilizes
jqfor robust parsing.\n - Capability inventory: The skill possesses the capability to modify local configuration files and create executable scripts within the user's
.claude/directory.\n - Sanitization: The script uses
jqto extract command strings safely and employs quoted shell variables during pattern matching to prevent shell injection risks.
Audit Metadata