The Agent Skills Directory

[COMMAND_EXECUTION]: The skill automates Electronic Design Automation (EDA) workflows by executing shell commands via make and subprocess. This is the intended functionality to interface with external toolchains like VCS, Verilator, and SpyGlass.
[SAFE]: Input validation is performed at the MCP server level using regular expressions (e.g., HDL_IDENTIFIER_RE, TEST_NAME_RE) to ensure that user-provided arguments like module names and test cases do not contain malicious shell metacharacters before they are passed to shell execution.
[SAFE]: The inclusion of a hygiene script (check_repo_hygiene.py) acts as a security guardrail by scanning for absolute paths and license server strings to prevent sensitive environment details from being committed to the repository.
[SAFE]: The skill implements a integrity verification system using SHA256 fingerprints to ensure that source files remain unchanged during long-running simulation or synthesis tasks, providing a tamper-evident execution loop.
[SAFE]: Resource access is scoped through a dedicated MCP runtime and configuration files, reducing the risk of unauthorized system-wide operations.

soc-build