The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: Fetches and executes the nvm installation script from an external repository using a high-risk piped execution pattern.
Evidence: curl -o- https://raw.githubusercontent.com/nvm-sh/nvm/v0.39.0/install.sh | bash in references/tool-installer.md.
[COMMAND_EXECUTION]: Performs global system modifications and destructive environment cleanup.
Evidence: Installs yarn and pnpm globally using npm install -g in references/tool-installer.md.
Evidence: Executes rm -rf node_modules and removes lock files (package-lock.json, yarn.lock, pnpm-lock.yaml) during the version alignment process in references/version-alignment.md.
Evidence: Uses dynamic command execution via find -execdir to run installation commands in subdirectories in references/version-alignment.md.
Evidence: Cleans the global npm cache using npm cache clean --force in references/version-alignment.md.
[PROMPT_INJECTION]: Significant attack surface for indirect prompt injection as the skill processes numerous untrusted project files.
Ingestion points: The skill reads package.json, source code (.tsx, .ts), CI/CD workflows, and configuration files across multiple steps (references/repository-inventory.md, references/config-analysis.md, references/cicd-analysis.md).
Boundary markers: Absent. There are no explicit instructions or delimiters used to prevent the agent from following instructions embedded in the analyzed code.
Capability inventory: The skill utilizes high-privilege tools including Bash, Write, Edit, and WebFetch, which could be exploited if malicious content is encountered in project data.
Sanitization: Absent. The skill analyzes raw file content from the repository without filtering or validation.

react-health-audit