The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process untrusted data from local directories (OKF bundles). This content (markdown and YAML) is interpolated into a generated website project.
Ingestion points: The assets/okf_site.py script scans and reads all markdown concept files and YAML frontmatter within the user-provided bundle directory.
Boundary markers: The skill does not implement specific delimiters or "ignore instructions" directives when processing bundle content, which could allow malicious instructions in the source files to influence the agent.
Capability inventory: The script performs file system operations including directory creation, deletion (shutil.rmtree), and file writing (open().write()) within the designated output path.
Sanitization: The script includes html.escape() calls when rendering metadata badges and extension keys to mitigate XSS and injection in the generated output.
[COMMAND_EXECUTION]: The skill workflow requires the agent to execute local Python scripts and standard Node.js build tools.
Evidence: Instructions in SKILL.md and README.md guide the agent to execute python3 assets/okf_site.py for inspection and generation, followed by npm install and npm run build to compile the generated Astro project.

okf-site-kit