The Agent Skills Directory

[METADATA_POISONING]: The skill metadata contains deceptive attribution. While the skill is authored by 'abhinavshrivastava950', the YAML frontmatter in SKILL.md explicitly lists 'vercel' as the author. Although the body of SKILL.md mentions the content is 'vendored from Vercel', the author field in the metadata is misleading.
Evidence: SKILL.md frontmatter contains metadata: author: vercel.
[INDIRECT_PROMPT_INJECTION]: The skill instructions direct the agent to ingest and refactor external React codebases, creating a surface for indirect prompt injection where malicious instructions embedded in a project's source code could influence the agent's behavior.
Ingestion points: External React codebase (specified as the target for refactoring tasks in SKILL.md).
Boundary markers: Absent. The skill does not provide delimiters or instructions to ignore embedded commands within the processed code.
Capability inventory: The skill is intended for use in tasks involving component refactoring, which typically requires file-system write access and potentially execution of build/test tools.
Sanitization: Absent. No guidelines are provided for escaping or validating the content of the components being refactored.

vercel-composition-patterns