The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill instructs the agent to download and execute shell scripts directly from external URLs by piping them to the bash interpreter. This allows for arbitrary code execution from remote sources without local integrity verification. Evidence: curl -L https://raw.githubusercontent.com/noir-lang/noirup/refs/heads/main/install | bash and curl -L https://raw.githubusercontent.com/AztecProtocol/aztec-packages/refs/heads/master/barretenberg/bbup/install | bash. Additionally, the skill suggests installing an external MCP server using npx @critesjosh/noir-mcp-server@latest, which involves fetching and executing code from an unverified third-party repository.
[COMMAND_EXECUTION]: The skill requires the execution of various command-line utilities, including nargo, bb, and shell-based installers, to manage the development toolchain and build process.
[EXTERNAL_DOWNLOADS]: The skill specifies multiple external dependencies that are fetched at runtime, including Git-based Noir libraries (noir-lang/poseidon, privacy-scaling-explorations/zk-kit.noir) and Node.js packages, which are not subject to standard security pinning or verification in the provided instructions.
[INDIRECT_PROMPT_INJECTION]: The skill processes user-supplied project files such as Nargo.toml, main.nr, and Prover.toml as ingestion points. Maliciously crafted data in these files could influence agent behavior during the compilation and proving phases.
Ingestion points: Nargo.toml, src/main.nr, Prover.toml.
Boundary markers: None implemented.
Capability inventory: Shell execution of nargo, bb, npm, and curl.
Sanitization: No sanitization or validation logic is defined for the ingestion of external project metadata or source code.

noir