debugging-code

Pass

Audited by Gen Agent Trust Hub on Mar 14, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill downloads the 'dap' CLI tool and its associated release metadata directly from the author's repository on GitHub.
  • Evidence: 'scripts/install-dap.sh' performs curl requests to 'api.github.com' and 'github.com/AlmogBaku/debug-skill'.
  • [EXTERNAL_DOWNLOADS]: The skill facilitates the installation of the 'js-debug' adapter from a trusted Microsoft repository.
  • Evidence: 'references/installing-debuggers.md' provides a command sequence to download and extract a tarball from 'github.com/microsoft/vscode-js-debug'.
  • [COMMAND_EXECUTION]: The installation script modifies user shell configuration files to update the system PATH.
  • Evidence: 'scripts/install-dap.sh' suggests appending an export command to shell-specific resource files like '.bashrc' or '.zshrc'.
  • [PROMPT_INJECTION]: The skill presents an indirect prompt injection surface by ingesting and presenting untrusted debugger data to the agent.
  • Ingestion points: Source code, local variable values, and call stack frames retrieved via the 'dap' tool in 'SKILL.md'.
  • Boundary markers: Absent; the system does not use specific delimiters or instructions to ignore potential commands embedded in the debugger's output.
  • Capability inventory: The agent possesses the 'Bash' tool with broad execution rights for the 'dap' command, as defined in 'SKILL.md'.
  • Sanitization: No evidence of output sanitization or validation exists for data returned from the debugger processes before it is integrated into the agent's context.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 14, 2026, 02:02 AM