The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill establishes a workflow for generating and bundling code into a self-contained HTML artifact. This creates a surface where the agent could inadvertently include malicious scripts or instructions in the project files during the development phase, which are then packaged into the final executable artifact.
Ingestion points: The skill instructions direct the agent to develop the artifact by editing source files in a project directory managed by the skill's scripts.
Capability inventory: The scripts/bundle-artifact.sh script executes shell commands to install dependencies, run the Parcel bundler, and use html-inline to generate the final file. scripts/init-artifact.sh uses sed, cat, and tar to set up the environment.
Boundary markers: No specific delimiters or safety warnings are implemented to distinguish between build instructions and the code being generated.
Sanitization: The skill does not perform sanitization or validation of the generated code before it is bundled into the HTML artifact.
[COMMAND_EXECUTION]: The provided shell scripts perform various system operations including file deletion (rm -rf), configuration generation (cat > file), and global package installation (npm install -g pnpm). These actions are standard for the intended purpose of setting up a local development environment but represent significant system interaction.

artifacts-builder