gsd-debug

Pass

Audited by Gen Agent Trust Hub on Apr 3, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill executes local shell commands and a Node.js script from a specific absolute path (/mnt/local-analysis/workspace-hub/.codex/get-shit-done/bin/gsd-tools.cjs). These commands are used to manage session state and resolve model configurations within the debugging framework.
  • [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection due to the way it handles untrusted data:
  • Ingestion points: Raw user input is captured via {{GSD_ARGS}} and subsequent symptom gathering steps (Expected behavior, Actual behavior, Error messages).
  • Boundary markers: There are no explicit delimiters or 'ignore' instructions surrounding the interpolated user content in the subagent prompts defined in references/debugger-prompt.md.
  • Capability inventory: The skill can spawn new agents (spawn_agent), read files (cat), and execute local scripts (node).
  • Sanitization: There is no evidence of input validation or sanitization before passing user-controlled text into the subagent's instructions. An attacker could provide malicious content within a simulated error message to influence the subagent's behavior.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 3, 2026, 04:33 AM