debug
Pass
Audited by Gen Agent Trust Hub on Aug 21, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [DYNAMIC_EXECUTION]: The skill generates and executes temporary scripts, reproduction scaffolds, and commands as part of the debugging workflow. While these are intended for diagnostic purposes, they involve the dynamic creation of executable logic based on the agent's analysis.
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process untrusted external data, including error messages, system logs, execution traces, and user-provided "real materials," which could potentially harbor adversarial instructions aimed at influencing the agent's behavior.
- Ingestion points: Workflow steps 1 and 2 involve reading external logs, traces, and diagnostic output.
- Boundary markers: The instructions do not define specific delimiters or instructions for the agent to ignore embedded commands within the processed data.
- Capability inventory: The agent has the capability to execute shell commands, run scripts, and write to the local repository via the
git-workflowskill. - Sanitization: No explicit sanitization or filtering of external content is mentioned in the workflow.
Audit Metadata