The Agent Skills Directory

[COMMAND_EXECUTION]: The script scripts/rfc.py utilizes subprocess.run to invoke system utilities like rsync for corpus synchronization and grep or ripgrep for full-text searching. All subprocess calls are implemented using argument lists without shell=True, which is a security best practice that prevents shell injection vulnerabilities.
[EXTERNAL_DOWNLOADS]: The skill downloads the RFC index and individual documents via HTTPS from www.rfc-editor.org. Additionally, the sync command allows the user to mirror the full 512 MB text corpus from rsync.rfc-editor.org. These are the official, well-known repositories for the IETF standards, and the skill includes warnings and confirmation prompts for the large-scale sync operation.
[DATA_EXFILTRATION]: No evidence of unauthorized data access or exfiltration. The skill's network activity is strictly limited to retrieving public technical documentation from authoritative sources. It uses a custom User-Agent that identifies the tool and its origin.
[PROMPT_INJECTION]: The markdown instructions in SKILL.md provide clear, technical guidelines for the AI agent to follow. They emphasize verification, correct citation, and checking for obsolete standards, without any attempts to override safety filters or manipulate the agent's core instructions.

rfc-lookup