The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill instructions require the agent to read and process content from external, user-provided repositories to validate them for catalog submission.
Ingestion points: The skill reads SKILL.md, all files in references/, and evals/ from the source repository being submitted (File: SKILL.md).
Boundary markers: There are no explicit delimiters or instructions to ignore embedded prompts within the external files being analyzed.
Capability inventory: The skill utilizes file system access, network access for Git operations, and execution of local build/validation scripts.
Sanitization: The instructions do not define a process for sanitizing or escaping the content of the external Markdown or YAML files before they are processed by the agent.
[COMMAND_EXECUTION]: The skill uses various command-line tools to perform its tasks, which is expected for a developer-oriented workflow.
Evidence: The skill executes skills-ref validate for validation, pnpm typecheck and pnpm build for website verification, and scripts/generate-og-image.py for asset generation. These tools are localized to the vendor's environment.

submit-context-skill