The Agent Skills Directory

[METADATA_POISONING]: The skill's primary documentation link and metadata author field reference a URL (jeffallan.github.io/claude-skills/skills/frontend/nextjs-developer/) that is flagged as malicious by multiple automated security scanners and is present on domain blacklists.
[INDIRECT_PROMPT_INJECTION]: The skill exhibits an attack surface for indirect prompt injection through its handling of external data sources.
Ingestion points: Data is ingested via formData in server actions, dynamic route params, and API request bodies as shown in references/server-actions.md and references/app-router.md.
Boundary markers: The provided code templates lack boundary markers or explicit instructions to prevent the model from interpreting data as instructions.
Capability inventory: The skill uses capabilities including file system modification (writeFile), network access (fetch), and shell command execution (next build).
Sanitization: Several templates show direct interpolation of unvalidated external data into metadata and UI components.
[DYNAMIC_EXECUTION]: In references/server-actions.md, the file upload example uses file.name directly in a path construction (join(process.cwd(), 'public', 'uploads', file.name)). This pattern is vulnerable to path traversal attacks, allowing a remote user to write files to arbitrary locations on the server by providing a filename with ../ sequences.
[DYNAMIC_EXECUTION]: The skill includes a pattern in references/server-components.md that uses dangerouslySetInnerHTML to render post.content. This bypasses standard security protections and could lead to Cross-Site Scripting (XSS) if the source content contains malicious scripts.
[PERSISTENCE]: The deployment documentation in references/deployment.md instructs users on how to establish persistence for the application using pm2 commands such as pm2 startup and pm2 save, which configures the software to launch automatically upon system boot.
[COMMAND_EXECUTION]: The core workflow and deployment references suggest executing shell commands (next build, npm run build, vercel) without providing guidance on verifying the integrity of the environment or the source code being built.

nextjs-developer