The Agent Skills Directory

[PROMPT_INJECTION]: As a code quality analysis tool, this skill processes external codebase content, creating a surface for indirect prompt injection. Maliciously crafted content within an analyzed project (such as comments or strings) could attempt to influence the agent's analysis results or refactoring suggestions.
Ingestion points: Untrusted source code enters the context through QualityOrchestrator.analyze_codebase(path=...) and ProactiveQualityScanner.scan(codebase=...) as defined in SKILL.md and modules/proactive-analysis.md.
Boundary markers: The documentation does not specify the use of delimiters or 'ignore' instructions to prevent the agent from obeying instructions embedded in the analyzed code.
Capability inventory: The skill utilizes Read, Grep, and Glob tools for inspection; it can modify files via auto_fix_issues and perform network operations via webhooks.
Sanitization: No explicit sanitization or filtering of codebase content is mentioned in the provided architectural documentation.
[COMMAND_EXECUTION]: The documentation and examples demonstrate the orchestration of various development tools (linters like ruff/eslint, formatters like black/prettier, and test frameworks like pytest/jest) using shell commands within CI/CD pipelines.
[DATA_EXFILTRATION]: The skill provides integration patterns for sending analysis reports and security findings to external services via Slack webhooks and GitHub Pull Request comments, as described in modules/integration-patterns.md.
[REMOTE_CODE_EXECUTION]: The 'auto-fix' and 'automated refactoring' features allow the agent to generate and apply code changes to the analyzed codebase. This capability allows for the generation of code that may be executed during subsequent CI/CD runs or by developers.
[NO_CODE]: No standalone executable script files (.py, .js, .sh) are included in the skill package; all logic and implementation details are provided as documentation and code snippets within Markdown files.

moai-foundation-quality