The Agent Skills Directory

[COMMAND_EXECUTION]: The skill documents the copier tool's ability to run shell tasks and migrations defined within templates.
Evidence: references/templating.md and references/update-migrations.md describe the _tasks and _migrations settings.
Security Context: The skill identifies that these actions require the --trust flag and warns users to only execute templates from trusted sources.
[EXTERNAL_DOWNLOADS]: The tool described in the skill interacts with remote Git repositories to fetch templates.
Evidence: SKILL.md and references/cli-api.md list GitHub, GitLab, and other Git providers as supported sources.
Security Context: This is a standard feature of the documented tool and involves well-known services.
[PROMPT_INJECTION]: The skill describes a workflow where the agent processes external template data that could contain malicious instructions (Indirect Prompt Injection surface).
Ingestion points: External Git repositories and template source paths defined in SKILL.md and references/cli-api.md.
Boundary markers: The documentation emphasizes the requirement for the --trust flag to execute unsafe template content.
Capability inventory: Templates can execute shell commands through _tasks and version migrations.
Sanitization: No automated sanitization is described; the tool relies on the operator to verify the template's source and content.

copier