The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes project-defined scripts such as npm run build and npm run start during the analysis and verification phases.
[COMMAND_EXECUTION]: The generated commands include the use of lsof -ti:3000 | xargs kill -9 to forcibly terminate processes running on specific ports.
[REMOTE_CODE_EXECUTION]: The skill's primary function is to generate new executable command files (markdown with bash scripts) in the .claude/commands/ directory, which are intended for later execution by the agent.
[EXTERNAL_DOWNLOADS]: The skill uses npx to fetch and run the lighthouse and playwright tools from the npm registry.
[DATA_EXFILTRATION]: The skill uses the GitHub CLI (gh pr create) to push code changes and performance reports (Lighthouse results) to remote repositories.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it ingests data from multiple project files to inform the generation of optimization plans and executable commands.
Ingestion points: package.json, webpack.config.*, vite.config.*, tsconfig.json, docs/regulation.md, server source code, and playwright.config.ts (SKILL.md).
Boundary markers: No specific delimiters or instructions are used to prevent the agent from following malicious instructions potentially embedded in the analyzed project files.
Capability inventory: The skill can execute arbitrary project scripts, manage processes, perform network requests via curl, and interact with Git and GitHub CLI tools.
Sanitization: Content extracted from project files is synthesized into new skills and commands without explicit sanitization or validation.

planning-wsh-improvements