agent-qa-debug-fix
Pass
Audited by Gen Agent Trust Hub on Aug 17, 2026
Risk Level: SAFECOMMAND_EXECUTIONDATA_EXFILTRATIONPROMPT_INJECTIONNO_CODE
Full Analysis
- [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it reads external data from 'agent_qa_get_run_logs' and 'agent_qa_get_run_artifact' which are not controlled by the agent. Maliciously crafted logs could attempt to influence the agent's logic during debugging or patching. Ingestion points: artifacts and logs via tools like 'agent_qa_get_run_logs' and 'agent_qa_get_run_artifact'. Boundary markers: Absent; there are no specific delimiters to separate untrusted data from the prompt context. Capability inventory: The skill can modify local source code and execute tests. Sanitization: Absent; the skill does not mention filtering or validating log content.
- [COMMAND_EXECUTION]: The skill instructs the agent to execute shell commands or tool calls to rerun tests, suites, and hooks for verification purposes as part of its primary workflow.
- [DATA_EXFILTRATION]: The skill uses 'dashboard REST APIs' to communicate with external servers when fetching QA run evidence if the standard MCP interface is unavailable.
- [NO_CODE]: The skill consists only of instructions in the SKILL.md file and does not include any accompanying scripts or executable files.
- [SAFE]: The instructions include safety boundaries requiring user confirmation for production-facing or destructive actions and explicitly forbid the exposure of credentials or sensitive application data found in artifacts and logs.
Audit Metadata