The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes various framework-specific CLI tools and package manager commands to scaffold projects and install dependencies. Evidence includes instructions to run npx create-next-app, bunx create-vite, uv init, cargo init, and go mod init (documented in references/stack-matrix.md).
[EXTERNAL_DOWNLOADS]: Fetches project templates and software packages from official registries (NPM, PyPI, Crates.io) and GitHub using standard development tools like npm, pnpm, bun, uv, cargo, and go. For example, it installs well-known packages like fastapi, next, and hono (documented in references/stack-matrix.md).
[PROMPT_INJECTION]: The skill inherits conventions from untrusted local files, which creates a surface for indirect prompt injection where instructions hidden in those files could influence agent behavior during bootstrapping.
Ingestion points: Reads CLAUDE.md, package.json, pyproject.toml, Cargo.toml, and lockfiles to extract architecture patterns and naming conventions (specified in references/convention-inheritance.md).
Boundary markers: No explicit boundary markers or instruction-guardrails are defined for the content read from these files.
Capability inventory: Performs shell command execution, remote package installation, and file system writes across multiple scripts (detailed in references/scaffolding-steps.md and references/stack-matrix.md).
Sanitization: No sanitization or validation logic is specified for the data extracted from the inherited repository files.

bootstrap