The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The script scripts/install-skills.py is designed to download and install external agent skills into the local environment using npx skills add. It specifically targets the GitHub repository wshobson/agents, which is an untrusted third-party source. Installing external skills allows remote instructions to influence agent behavior and execution.
[COMMAND_EXECUTION]: The skill uses subprocess.run in scripts/create.py and scripts/install-skills.py to execute system-level commands. This includes running package managers like bun and uv to scaffold project structures and the npx skills utility to install external extensions.
[EXTERNAL_DOWNLOADS]: The skill references and downloads content from several external sources. While some references target well-known organizations like Anthropics, it also initiates downloads from unverified third-party repositories (wshobson/agents).
[INDIRECT_PROMPT_INJECTION]: The skill establishes an attack surface for indirect injection by scaffolding a workflow that automatically fetches and processes an OpenAPI schema from a local development server (localhost:8000/openapi.json).
Ingestion points: scripts/create.py (generated generate-api script in package.json).
Boundary markers: None present in the generated code or instructions.
Capability inventory: subprocess.run and subprocess.call across multiple management scripts.
Sanitization: No validation or sanitization of the fetched schema is implemented before it is processed by code generation tools.

init-app-stack