The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill instructions utilize the mops package manager, which installs software components from remote sources such as GitHub and package registries.
[COMMAND_EXECUTION]: The skill requires the agent to execute shell commands using the mops toolchain for dependency management and project building (mops add, mops install, mops build).
[REMOTE_CODE_EXECUTION]: Dependency installation and compilation processes involve downloading and executing third-party code, which carries inherent risks of running untrusted logic.
[PROMPT_INJECTION]: Automated scanning flagged the documentation site https://docs.mops.one/ as a phishing threat; the skill directs the agent to consult this site for troubleshooting, which could expose the agent to malicious content.
[PROMPT_INJECTION]: The skill possesses a surface for indirect prompt injection via its processing of untrusted code.
Ingestion points: Reads Motoko source files (.mo), mops.toml, and mops.lock files from the local environment.
Boundary markers: The instructions lack markers or specific instructions to distinguish code logic from potential embedded natural language instructions.
Capability inventory: The skill uses tools with file system modification (mops check --fix), network access (mops add), and code compilation (mops build) capabilities.
Sanitization: The skill does not define any sanitization, validation, or filtering mechanisms for the content of the input files.

writing-motoko