The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes user-provided smart contract files (.sol, .cairo, .rs, .move) to identify patterns and integrate library components. This ingestion of untrusted data creates a surface for indirect prompt injection attacks.
Ingestion points: Project contracts are read using Glob patterns and Read tools during the pattern discovery phase.
Boundary markers: The skill does not explicitly define delimiters or instructions for the agent to ignore potentially malicious content within these files.
Capability inventory: The agent is equipped with the Edit tool for modifying files and bash for executing command-line operations.
Sanitization: No specific validation or sanitization of the ingested contract code is mentioned.
[COMMAND_EXECUTION]: The skill instructions direct the agent to use bash for various tasks, including running the OpenZeppelin CLI, performing file diffs, and cloning repositories. These commands are integral to the developer workflow described.
[EXTERNAL_DOWNLOADS]: The skill fetches code and documentation from official OpenZeppelin GitHub repositories and utilizes the NPM registry to run the @openzeppelin/contracts-cli package. These operations target trusted vendor-managed infrastructure.
[DYNAMIC_EXECUTION]: The skill employs a 'generate-compare-apply' workflow that involves running the @openzeppelin/contracts-cli package at runtime to generate baseline and variant contracts for comparison.

develop-secure-contracts