The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes a Python script to interact with the 51CTO API. The inputs (title, content, tags) are passed as command-line arguments. The script uses standard libraries and implements validation for content size and structure before execution.- [CREDENTIALS_SAFE]: The skill uses a 'Bring Your Own Cookie' (BYOC) model where credentials are provided via an environment variable CTO51_COOKIES. The script explicitly redacts CSRF tokens from error messages and logs to prevent accidental exposure and implements strict domain-matching for cookies to ensure they are only sent to legitimate 51CTO endpoints.- [DATA_EXFILTRATION]: Network activity is restricted to blog.51cto.com. The script uses a custom HTTP handler that disables redirects to prevent sensitive headers or cookies from being leaked to third-party domains if the API issues a redirect.- [PROMPT_INJECTION]: The skill includes instructions to the agent to always perform a dry run first and wait for explicit user confirmation before creating a draft. It also mandates that the agent tell the user the article is only a draft and not live.- [SAFE]: The Python script is written using only standard libraries (urllib, json, argparse, etc.), meaning it requires no external third-party dependencies that could introduce supply-chain risks.

cto51