The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill instructions involve processing project-specific files such as mops.toml and Motoko source files (.mo). While this constitutes a data ingestion surface, the operations described are standard for the tool's primary purpose of project management and compilation.
Ingestion points: Reads configuration from mops.toml and source code from Motoko files.
Boundary markers: Standard file parsing by the mops CLI tool.
Capability inventory: Invokes shell commands for the mops CLI, the Motoko compiler (moc), and related tools like pocket-ic and lintoko.
Sanitization: Relies on the internal validation and sanitization logic of the underlying mops toolchain.
[DATA_EXPOSURE_AND_EXFILTRATION]: The guide mentions using the GITHUB_TOKEN environment variable to authenticate with the GitHub API for dependency resolution. This is documented as a standard practice to avoid API rate limits and does not involve exfiltration of sensitive data.
[COMMAND_EXECUTION]: The skill provides numerous examples of CLI commands (e.g., mops install, mops build, mops toolchain use). All commands are standard operations for the mops development environment and do not include arbitrary or dangerous shell execution patterns.

mops-cli