The Agent Skills Directory

[SAFE]: The skill provides a structured 10-stage workflow for research and manuscript preparation, prioritizing user confirmation at every stage.
[SAFE]: Includes a dedicated integrity_verification_agent that performs 100% verification of references and data claims using automated searches to prevent the inclusion of fabricated content or AI hallucinations.
[SAFE]: Implements mandatory checkpoints for integrity and review boundaries, ensuring that papers cannot proceed to publication without passing rigorous verification gates.
[SAFE]: Uses industry-standard tools like pandoc and tectonic for document conversion and PDF compilation. These commands are executed on agent-managed content within the intended scope of academic formatting.
[SAFE]: The skill demonstrates best practices by enforcing structured data handoffs between stages (using defined schemas) and maintaining a comprehensive state audit trail via the state_tracker_agent.
[SAFE]: No evidence of prompt injection, obfuscation, or unauthorized access to sensitive file paths was found across the 13 files analyzed.

academic-pipeline