The Agent Skills Directory

[SAFE]: The skill is designed for the localized cleanup of specific project scaffolding and does not exhibit malicious patterns.
[DATA_EXF1LTRATION]: The skill demonstrates security-positive behavior by using sed to redact sensitive values in environment files (e.g., .env, .env.local) before the agent processes or outputs the findings.
[COMMAND_EXECUTION]: The commands provided (grep, npm uninstall, mktemp, diff) are standard administrative and development tasks performed on the local filesystem. They are used exclusively for finding and removing specific strings or packages related to the skill's purpose.
[INDIRECT_PROMPT_INJECTION]: The skill processes external, potentially untrusted project data (source code, README files, package manifests).
Ingestion points: Files in the project root and src/ directory are scanned via grep.
Boundary markers: None explicitly defined for interpolated data, though the skill focuses on specific string matches.
Capability inventory: File system scanning, local package removal (npm uninstall), and project build execution (npm run build).
Sanitization: Values from environment variables are redacted using sed regex to prevent sensitive data exposure to the agent.

lovable-cleanup