The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill facilitates reading and acting upon local configuration files which may contain untrusted data provided by an external source.
Ingestion points: The skill reads mise.toml, .tool-versions, .nvmrc, and .python-version files from the project directory to configure the environment (SKILL.md).
Boundary markers: None identified; the agent is instructed to directly process these files to determine tool versions and environment settings.
Capability inventory: The skill possesses the capability to execute shell commands via the Bash tool, perform network downloads via mise install, and run arbitrary scripts via postinstall hooks (SKILL.md).
Sanitization: No explicit sanitization or validation of the version strings or hook scripts read from these external configuration files is described.
[COMMAND_EXECUTION]: The skill demonstrates the execution of shell commands for tool management and lifecycle hooks.
Evidence: The skill utilizes postinstall hooks in mise.toml to execute shell commands such as corepack enable and npm install -g npm@latest (SKILL.md).
[DYNAMIC_EXECUTION]: The skill uses configuration hooks to execute dynamically defined scripts during the tool installation process.
Evidence: Example configurations in mise.toml show the use of the postinstall field to run multi-line shell scripts after tool installation (SKILL.md).
[EXTERNAL_DOWNLOADS]: The skill downloads and installs development tools from remote registries and repositories.
Evidence: The skill instructions include installing packages from the npm, cargo, and go registries, as well as cloning and installing from Git repositories (e.g., git:https://github.com/org/tool.git) (SKILL.md).

mise-tool-management