slate-ar-quality
Pass
Audited by Gen Agent Trust Hub on Jul 2, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes project-specific CLI tools (autoresearch-cli, slate-ar) to perform automated research and quality analysis within a restricted temporary directory (.tmp/slate-v2).
- [PROMPT_INJECTION]: The skill has a surface for indirect prompt injection as it interpolates user-provided $ARGUMENTS into shell command strings. This is mitigated by the use of double-quoting for parameters and an explicit instruction to the agent to require human review before applying any generated changes. • Ingestion points: $ARGUMENTS mapping to --slug and --goal parameters in SKILL.md. • Boundary markers: Parameters are enclosed in double quotes in the command template. • Capability inventory: Ability to execute research and propose code changes via the CLI tool. • Sanitization: Explicit instruction for manual inspection before executing commands with the --apply flag.
Audit Metadata