The Agent Skills Directory

[Category 8: Indirect Prompt Injection] (SAFE): The skill provides code patterns for fetching and processing data from external sources (APIs, files) which are common in I/O optimization. While these present an ingestion surface, they are passive instructional examples. • Ingestion points: fetch_all_apis(urls) in conc-asyncio-for-io.md, fetch_all_urls(urls) in io-semaphore.md, and process_large_file(filepath) in mem-generators.md. • Boundary markers: Absent in code snippets. • Capability inventory: Network requests (aiohttp) and file reads (aiofiles, open). • Sanitization: Absent, as common in high-level coding snippets.
[Category 4: External Dependencies] (SAFE): The skill recommends several well-known and standard Python libraries for optimization (e.g., uvloop, aiohttp, asyncpg). These are referred to for educational purposes and are not installed or executed by the skill itself.
[Category 10: Dynamic Execution] (SAFE): Use of functools.partial and @lru_cache in examples are standard Python idioms for performance and do not involve untrusted code execution.

python