The Agent Skills Directory

[COMMAND_EXECUTION]: The skill is designed to author bash scripts that execute system commands for environment management and infrastructure setup.
Evidence: The template.sh file includes functions for modifying local files (write_env) and interacting with the GitHub CLI (set_secret, set_var). The instructions in SKILL.md guide the creation of scripts that the user is encouraged to run locally.
[INDIRECT_PROMPT_INJECTION]: The skill identifies a surface for indirect prompt injection by reading repository files to determine its configuration steps.
Ingestion points: According to SKILL.md, the skill reads .env files, README.md, docker-compose files, and .github/workflows/ scripts to map required secrets and variables.
Boundary markers: No specific delimiters or "ignore" instructions are used when parsing these external files to prevent malicious content from influencing the generated script.
Capability inventory: The generated scripts can write to the filesystem, perform network operations via the gh tool, and open arbitrary URLs in a web browser.
Sanitization: The skill does not explicitly define sanitization logic for data ingested from the repository files before incorporating it into the bash output.

wizard