rebuttal
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill implements a structured workflow for handling conference submission rebuttals. It tracks reviewer concerns, structures responses based on venue rules, and enforces boundaries to ensure that no statements or figures are fabricated.
- [COMMAND_EXECUTION]: The skill uses conditional workflow prompts to call external testing backends (
mcp__codex__codex,mcp__manual_review__review) to simulate peer reviews. These configurations are well-contained inside structural constraints and present no command injection risk. - [INDIRECT_PROMPT_INJECTION]: The skill ingests raw peer review texts (
rebuttal/REVIEWS_RAW.md). It possesses specific boundary rules, documentation patterns, and safety linting gates (Provenance gate, Commitment gate, Coverage gate) that prevent external text input from overriding agent logic.
Audit Metadata