The Agent Skills Directory

[COMMAND_EXECUTION]: The script scripts/measure-build.sh identifies the build command from the project's package.json and executes it using the detected package manager (e.g., npm run build). While this is the intended primary purpose of the skill, executing user-defined scripts from a project directory is a known capability surface.
[INDIRECT_PROMPT_INJECTION]: The skill ingests and processes untrusted data from the local environment, creating a potential surface for indirect prompt injection.
Ingestion points: The scripts detect-nextjs.sh, check-upgrade-path.sh, and measure-build.sh read package.json, lock files, and build log outputs.
Boundary markers: The skill does not implement specific delimiters or warnings to the agent to ignore instructions that might be embedded within the project files it reads.
Capability inventory: The skill possesses Bash, Read, Write, and Edit permissions, allowing it to modify the project based on the data it parses.
Sanitization: The scripts use node -e to process strings and JSON, which provides some structural validation, but they do not filter the content for natural language instructions that might target the LLM agent.

nextjs-upgrade