The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests data from local component source files to generate documentation, exposing an attack surface for indirect prompt injection.
Ingestion points: The agent reads file content from path patterns like src/components/ui/[component-name]/index.tsx and packages/gluestack-core/src/[component-name]/creator/index.tsx in Phase 2.
Boundary markers: There are no explicit instructions or delimiters used to ensure the agent ignores malicious instructions that might be embedded in the component code or comments.
Capability inventory: The skill allows the agent to write new MDX and Handlebars files, update sidebar.json, and execute shell commands such as yarn dev and git commit.
Sanitization: The skill does not implement sanitization or validation of the ingested code before using it to plan the documentation structure or generate summaries.
[COMMAND_EXECUTION]: The skill incorporates shell command execution as part of the documentation and development workflow.
The agent is instructed to run yarn dev within the apps/website directory to preview documentation.
The skill defines a dynamic installation command in generated documentation using process.env.NEXT_PUBLIC_GLUESTACK_COMMAND, defaulting to the vendor-provided npx gluestack-ui tool.

create-docs