The Agent Skills Directory

[SAFE]: The skill consists of documentation and reference material for performance optimization in the Bun runtime. It provides instructions for using official profiling flags and standard APIs.\n- [INDIRECT_PROMPT_INJECTION]: The skill describes how to handle network requests using Bun.serve, which introduces a surface for untrusted data ingestion. While it warns users to validate request-derived paths, it does not provide explicit instruction delimiters.\n
Ingestion points: Bun.serve in references/native-apis-and-concurrency.md processes the req object.\n
Boundary markers: Absent in the provided snippets.\n
Capability inventory: Uses Bun.file (read), Bun.write (write), and Worker (execution) in references/native-apis-and-concurrency.md.\n
Sanitization: The skill explicitly recommends path validation before file access in references/native-apis-and-concurrency.md.

bun-performance