The Agent Skills Directory

[SAFE]: The skill defines a structured workflow for repository exploration and technical fact-verification without introducing any executable code or external dependencies. It explicitly restricts the agent to a read-only role, stating that the subagent should perform "no design, no questions to the user, no code."
[SAFE]: No evidence of prompt injection or adversarial techniques was detected. The instructions focus on objective fact-gathering from a codebase to verify technical claims made by a user.
[SAFE]: The skill does not contain any network operations, sensitive file access (beyond the codebase itself for analysis), or obfuscated content. It operates entirely within the scope of code analysis.
[SAFE]: Although the skill processes repository data (an indirect prompt injection surface), it contains no capabilities for writing to the file system, executing commands, or sending data over a network, effectively neutralizing the risk of data exfiltration or system modification.

explore