The Agent Skills Directory

[COMMAND_EXECUTION]: The skill includes local bash and Python scripts, specifically scripts/count_limits.sh and scripts/init_rebuttal_pack.py. These scripts facilitate project scaffolding and length verification for rebuttal drafts. They utilize standard system utilities like sed, wc, and bc to perform text processing on local files, which is consistent with the skill's stated purpose and does not pose a security risk.
[EXTERNAL_DOWNLOADS]: Extensive references are made to official documentation and guidelines from reputable academic conferences and platforms (e.g., OpenReview, NeurIPS, ICLR, CVPR, ICML). These informational links are used to verify venue-specific policies and do not involve the download or execution of remote code.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to process external reviewer comments and feedback. To mitigate potential risks from untrusted data ingestion, the workflow incorporates structured 'input contracts' and 'provenance gates.' These measures ensure that drafted responses are grounded in verifiable manuscript evidence, effectively preventing the agent from following unintended instructions embedded in reviewer text.

research-rebuttal