The Agent Skills Directory

[SAFE]: The skill is entirely instructional and does not include any executable code, scripts, binary files, or network operations. It uses frontmatter to disable tool invocation, further limiting its capabilities to pure text processing.
[INDIRECT_PROMPT_INJECTION]: The skill directs the agent to incorporate information from CONTEXT.md and CONTEXT-MAP.md. While these constitute ingestion points for external data, the skill possesses no tools or permissions (command execution, network access) to perform harmful actions.
[NO_CODE]: No scripts or binaries are provided with this skill; it consists solely of a markdown prompt and a configuration file.

wait-what