The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes various external data sources, including interview transcripts, literature databases, and quantitative datasets, which represent a potential surface for indirect prompt injection. However, the skill mitigates this risk by using structured state management (project.yaml and progress.yaml), explicit dispatch templates that scope sub-agent tasks, and strict phase-based execution logic that limits the influence of external content on the agent's core instructions.
[SAFE]: The skill follows security best practices for research toolkits, including the use of git for version history rather than duplicate files, and standard system tools like grep for verification. All sub-skill invocations and file operations are aligned with the primary purpose of research orchestration, with no evidence of obfuscation, unauthorized network operations, or credential harvesting.

research-coordinator