The Agent Skills Directory

[COMMAND_EXECUTION]: The skill includes utility scripts (inspect-description.ts, setup-codex-icons.ts) designed to run in the local environment using the Bun runtime.
Evidence: These scripts are located in the scripts/ directory and are used to automate the validation of skill descriptions and the generation of icons for the Codex platform.
Capabilities: The scripts utilize standard modules (node:fs, node:path, node:os) to perform file system operations, such as reading skill frontmatter and writing asset files, which are appropriate for their management functions.
[INDIRECT_PROMPT_INJECTION]: The skill serves as an auditor for other skill files, creating a surface where content from external skills is processed and validated.
Ingestion points: The validation and setup scripts read data from SKILL.md and openai.yaml files within local directories specified by the user or identified via directory traversal.
Boundary markers: The scripts expect content to be structured with YAML frontmatter delimited by --- and follow specific schema keys.
Capability inventory: The scripts are capable of reading and writing to the local file system within the user's skill repository to store generated SVG icons and update configuration files.
Sanitization: Input validation is performed in inspect-description.ts, including character length constraints (50-300 characters for descriptions) and regex-based validation for skill names to prevent malformed metadata from being processed.

skill-manager