gsd-debug
Pass
Audited by Gen Agent Trust Hub on Oct 4, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted user input through the $ARGUMENTS block which is used to define debugging tasks and interact with subagents.
- Ingestion points: The block in SKILL.md captures all user-supplied text.
- Boundary markers: The instructions attempt to wrap user input in tags and state it is data, not instructions, though this does not technically prevent injection.
- Capability inventory: The skill has access to Bash, Write, and Agent (subagent spawning) tools.
- Sanitization: No explicit sanitization, escaping, or schema validation is applied to the user input before it is used by the orchestrator or subagents.
- [COMMAND_EXECUTION]: The skill directs the agent to execute shell commands to manage its internal state.
- Evidence: Use of the Bash tool to run 'ls .planning/debug/*.md 2>/dev/null | grep -v resolved | head -5' for session tracking.
Audit Metadata