История аудитов

The audit found many Markdown backtick false positives, but several bash workflow blocks remain real external-command surfaces. The highest confirmed risk is authorization to write ~/.claude settings, plus an overbroad instruction block that can weaken normal instruction hierarchy.

The audit found many Markdown backtick false positives, but several bash workflow blocks remain real external-command surfaces. The highest confirmed risk is authorization to write ~/.claude settings, plus an overbroad instruction block that can weaken normal instruction hierarchy.

Static findings are mostly command examples and Markdown patterns, not installed executable code. The real risk is operational: the skill instructs agents to run shell and GitHub CLI commands, edit repository files, and allows writes to hidden Claude configuration, so publication should include a warning.

Documentation-only skill containing markdown examples of shell commands for requirements management. All detected patterns (external_commands, filesystem access, version strings) are false positives - the file contains documentation with example bash scripts in code blocks, not executable code. No malicious patterns, command injection vectors, or dangerous operations found.

This is a pure documentation skill containing only markdown guidelines for requirements management. The static analyzer incorrectly flagged documentation syntax (markdown backticks, version strings, bash examples in code blocks) as security issues. No executable code, network calls, or malicious patterns exist. All 73 findings are false positives.

This is a pure documentation skill containing only markdown guidelines for requirements management. The static analyzer incorrectly flagged documentation syntax (markdown backticks, version strings, bash examples in code blocks) as security issues. No executable code, network calls, or malicious patterns exist. All 73 findings are false positives.

This is a pure prompt-based documentation skill containing only markdown guidelines and templates for requirements management. No executable code, no network calls, no filesystem manipulation beyond normal AI assistant operations. The bash examples shown are illustrative documentation only.

This is a pure prompt-based documentation skill containing only markdown guidelines and templates for requirements management. No executable code, no network calls, no filesystem manipulation beyond normal AI assistant operations. The bash examples shown are illustrative documentation only.

This is a pure prompt-based documentation skill containing only markdown guidelines and templates for requirements management. No executable code, no network calls, no filesystem manipulation beyond normal AI assistant operations. The bash examples shown are illustrative documentation only.