The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill instructs the agent to install toolchains by piping remote shell scripts directly into bash. This pattern is highly susceptible to supply chain attacks and remote code execution if the source is compromised.
Evidence: curl -L https://raw.githubusercontent.com/noir-lang/noirup/refs/heads/main/install | bash in SKILL.md.
Evidence: curl -L https://raw.githubusercontent.com/AztecProtocol/aztec-packages/refs/heads/master/barretenberg/bbup/install | bash in SKILL.md.
[COMMAND_EXECUTION]: The skill relies extensively on shell commands for compiling circuits, generating proofs, and managing project infrastructure.
Evidence: Execution of nargo compile, nargo execute, bb write_vk, and bb write_solidity_verifier for the build pipeline.
Evidence: Local testing using bb prove and bb verify.
[EXTERNAL_DOWNLOADS]: The skill fetches various dependencies and tools from GitHub repositories and package registries.
Evidence: GitHub dependencies in Nargo.toml for poseidon and zk-kit.noir.
Evidence: Installation of MCP servers and Node.js packages like @noir-lang/noir_js and @aztec/bb.js.
[DYNAMIC_EXECUTION]: The skill uses the NoirJS library to dynamically load and execute circuit bytecode within a browser or Node.js environment.
Evidence: const backend = new UltraHonkBackend(circuit.bytecode, bb); followed by await noir.execute(inputs);.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data which could be manipulated to influence the agent's behavior.
Ingestion points: Circuit artifacts (.json files) loaded via fetch, prover inputs from Prover.toml, and Merkle tree siblings derived from blockchain events.
Boundary markers: No explicit delimiters or instructions are provided to the agent to ignore potentially malicious embedded content in these inputs.
Capability inventory: The agent has access to shell execution (nargo, bb), file writing to the project directory, and network access via package managers and curl.
Sanitization: No sanitization or schema validation steps are described for the circuit artifacts or external Merkle data.

noir