qemu-build
Pass
Audited by Gen Agent Trust Hub on Jul 31, 2026
Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [PROMPT_INJECTION]: The skill instructions create a surface for indirect prompt injection by having the agent process data from untrusted build artifacts.\n
- Ingestion points: Processes
config.log,build.ninja,meson-log.txt, andintro-buildoptions.json(SKILL.md).\n - Boundary markers: Absent. There are no instructions to the agent to treat this data as untrusted or to ignore embedded instructions.\n
- Capability inventory: Execution of
configureandninjacommands (SKILL.md).\n - Sanitization: Absent. The skill directs the agent to "reconstruct the old command from /config.log" without specifying validation steps.\n- [COMMAND_EXECUTION]: The skill instructions involve executing local shell commands for build configuration and project management.\n
- Evidence: Executing
configure,ninja, andgit rev-parsewithin the QEMU workspace (SKILL.md).
Audit Metadata