The Agent Skills Directory

[SAFE]: The skill is entirely prompt-based and does not contain any executable scripts, binary files, or configuration for external tools (MCP, hooks). It performs its task by applying logic described in markdown instructions and reference files.
[EXTERNAL_DOWNLOADS]: The skill's metadata and documentation reference official legal guidelines and rules from well-known professional organizations including the International Bar Association (IBA), International Chamber of Commerce (ICC), LCIA, and the World Bank (ICSID). These references are used for provenance and verification of the legal instructions provided.
[DATA_EXFILTRATION]: There is a strong emphasis on data privacy and legal privilege. The skill implements a 'Privilege Gate'—a mandatory hard stop requiring user confirmation before any data is processed—and explicitly instructs the model not to send data to cloud endpoints without explicit consent.
[PROMPT_INJECTION]: The instructions are highly structured and focused on procedural accuracy. There are no attempts to override system safety filters or engage in role-play that would bypass agent constraints.
[COMMAND_EXECUTION]: No shell commands or system-level operations are present in the instruction set. Mentions of CLI tools (e.g., curl) are contained within documentation explaining how the author verified legal citations offline and are not part of the active skill workflow.

redfern-schedule