autopilot
Pass
Audited by Gen Agent Trust Hub on Jul 29, 2026
Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [PROMPT_INJECTION]: The skill implements an autonomous delivery loop that ingests untrusted data from the user and the codebase, creating a surface for indirect prompt injection.
- Ingestion points: The process begins with a user-supplied activation prompt and involves reading codebase artifacts using tools like Read, Grep, Glob, and Bash.
- Boundary markers: The skill does not define clear delimiters or provide instructions to the agent to disregard instructions embedded in the processed data.
- Capability inventory: Through the ultragoal skill, the orchestrator has the capability to write files, execute build/lint commands, and run tests. It also uses Bash for environment inspection.
- Sanitization: No specific filtering or sanitization of the input data is documented before it is utilized in the implementation and verification phases.
- [COMMAND_EXECUTION]: The skill describes using system tools such as read-only Bash, Grep, Glob, and Read for repository analysis and state management during the pre-context intake phase.
Audit Metadata