The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill instructs the agent on using tools that ingest external file content, creating an attack surface where untrusted data could influence agent behavior.
Ingestion points: The skill describes usage of Read and Grep tools to fetch file content (SKILL.md).
Boundary markers: The instructions do not specify the use of delimiters or warnings to ignore embedded instructions in the ingested data.
Capability inventory: The skill utilizes Bash for shell execution and Edit for file modifications, which are powerful capabilities if triggered by malicious data.
Sanitization: No explicit content sanitization or validation methods are described for handling external tool outputs.
[DYNAMIC_EXECUTION]: To optimize performance, the skill recommends that the agent generate and execute its own scripts (Node.js or Python) for bulk operations, which is a form of dynamic code generation and execution.
[COMMAND_EXECUTION]: The skill provides patterns for using shell execution tools (Bash, run_in_terminal) to perform connectivity checks, run build tools like pnpm, and execute dynamically created scripts.
[EXTERNAL_DOWNLOADS]: The skill references grounding materials from well-known services (OpenAI, Anthropic) and the author's own GitHub repositories. These references are documented for informational purposes and are consistent with the skill's technical domain.

tool-call-strategy