The Agent Skills Directory

[Command Execution] (LOW): The script scripts/with_server.py facilitates running local server commands for testing purposes. It follows security best practices by using shell=False and a custom validator that explicitly blocks shell operators like '&&', '||', and '|' to prevent command injection (CWE-78).
[Indirect Prompt Injection] (LOW): The skill dynamically ingests patterns and best practices from an external 'Context7 MCP' service. This creates a surface where poisoned external documentation could theoretically influence the agent's code analysis results.
Ingestion points: context7_client.get_library_docs calls throughout modules/performance-optimization/ai-optimization.md, modules/automated-code-review/context7-integration.md, and modules/automated-code-review/trust5-framework.md.
Boundary markers: None identified; the skill assumes the retrieved patterns are trusted.
Capability inventory: Access to subprocess.Popen and subprocess.run via scripts/with_server.py.
Sanitization: No specific filtering of the content returned by the Context7 client is implemented.
[Dynamic Execution] (SAFE): The skill uses Python's ast module to parse and analyze source code for quality metrics and bug detection. This is performed for static analysis purposes and does not execute the code being analyzed.

moai-workflow-testing