The Agent Skills Directory

[DYNAMIC_EXECUTION]: The skill instructs the agent to generate and execute custom scripts at runtime.
The documentation guides the agent to create standalone Python (.py) or Shell (.sh) files and execute them using execute_shell_command or run_tool_batch to automate data processing.
It encourages the use of inline command execution, such as python3 -c "...", to replace manual agent judgment with rule-based automation.
[INDIRECT_PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection as it transforms conversational data into executable instructions.
Ingestion points: The focus, description, and logic of the new skill are derived directly from the "current conversation, workflow, or troubleshooting path" as seen in SKILL.md.
Boundary markers: The instructions do not define delimiters or specific instructions for processing potentially malicious data embedded in the conversation history.
Capability inventory: The skill has access to powerful tools including materialize_skill, execute_shell_command, run_tool_batch, and spawn_subagent.
Sanitization: The skill relies on an external security scanner within the materialize_skill tool rather than implementing its own sanitization or validation of the input data.
[COMMAND_EXECUTION]: The skill relies on shell command execution for its primary automation logic.
It instructs the agent to use execute_shell_command to run scripts for content extraction, filtering, and format conversion.
[EXTERNAL_DOWNLOADS]: The skill documentation references a public repository from a known organization.
A reference to the Anthropic skills GitHub repository is included in the documentation comments as inspiration for the skill creator portion.

make-skill