The Agent Skills Directory

[DATA_EXFILTRATION]: The debug-tricks.md file provides instructions for the agent to interact with a local /_next/mcp development endpoint. Specific tools mentioned, such as get_project_metadata and get_logs, are documented to return absolute filesystem paths (e.g., projectPath and log file locations), revealing internal directory structures and environment details to the agent.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to have the agent analyze external, untrusted data including project source code, error logs (next-development.log), and JSON-RPC tool outputs.
Ingestion points: Project source files, server log files, and responses from the local /_next/mcp dev server endpoint.
Boundary markers: Absent. The instructions do not provide delimiters or "ignore embedded instructions" warnings for the agent when it processes project data or log content.
Capability inventory: The skill includes several shell execution capabilities, such as npx @next/codemod, next build, and pm2 process management.
Sanitization: Absent. There are no instructions for the agent to sanitize, validate, or escape the content retrieved from the development environment before interpreting it or using it to formulate commands.
[COMMAND_EXECUTION]: The skill instructs the agent to execute various shell commands such as npx @next/codemod, next build --debug-build-paths, and pm2. While these are standard developer tools, they represent a significant capability that could be abused if the agent's logic is subverted by malicious content discovered during its analysis of project files or logs.

next-best-practices