The Agent Skills Directory

[REMOTE_CODE_EXECUTION] (CRITICAL): The documentation in examples/ci-workflow.yml includes a pattern for downloading and immediately executing a shell script from the internet using curl -fsSL https://moonrepo.dev/install/proto.sh | bash. This is a classic remote code execution vector.
[EXTERNAL_DOWNLOADS] (HIGH): The skill documentation in references/workspace-config.md describes a feature for loading remote WASM extensions from arbitrary URLs, such as plugin: "https://example.com/migrate-nx.wasm". Executing unverified remote binaries is a high-risk activity.
[COMMAND_EXECUTION] (MEDIUM): The skill is designed to manage and execute shell commands via the moon binary. As documented in references/v2-migration.md, the script field in moon.yml explicitly enables shell features like pipes, redirects, and environment expansion, allowing for arbitrary command execution on the host system.
[PROMPT_INJECTION] (LOW): There is a potential for Indirect Prompt Injection because the skill generates configuration files (moon.yml) containing shell scripts based on user requests.
Ingestion points: SKILL.md identifies triggers like "create moon tasks" or "add moon project" which process user-provided task descriptions.
Boundary markers: None are present in the provided templates or instructions to prevent the agent from obeying instructions embedded in user input when generating task scripts.
Capability inventory: The skill enables full shell execution via the script field in moon.yml (referenced in references/v2-migration.md) and task execution via moon run (referenced in references/cli-reference.md).
Sanitization: No evidence of sanitization or escaping of user input before interpolation into command or script fields.

moon