The Agent Skills Directory

[DYNAMIC_EXECUTION]: The skill documents the use of +exec and +exec_replace attributes within markdown code blocks, which allow for the execution of shell commands when the presentation is loaded or a key is pressed. These features are documented as opt-in, requiring specific command-line flags or configuration changes to function.
[INDIRECT_PROMPT_INJECTION]: The skill describes a process for handling external markdown files that could contain embedded commands intended for execution.
Ingestion points: Markdown files processed by the presenterm tool as described in SKILL.md.
Boundary markers: No specific boundary markers or sanitization requirements are defined for the markdown slide content.
Capability inventory: Shell command execution via the presenterm CLI as described in the code block attributes section.
Sanitization: The tool defaults to disabled execution, requiring explicit user consent via the -x or -X flags, or the enable_snippet_execution configuration setting.
[COMMAND_EXECUTION]: The documentation includes examples of shell commands used to run presentations, export them to PDF, and manage Kubernetes resources via kubectl within code blocks.
[EXTERNAL_DOWNLOADS]: The instructions suggest installing several well-known external tools and libraries for diagram rendering, including @mermaid-js/mermaid-cli from npm, d2, typst, and pandoc.

presenterm