The Agent Skills Directory

[SAFE]: The skill consists exclusively of Markdown documentation and instructional references. There are no scripts (.py, .js, .sh), binaries, or configuration files that would enable automated execution or external interactions.
[PROMPT_INJECTION]: The skill includes instructions to prioritize the user's current task and source evidence over the skill's own patterns (e.g., 'Michael's current instruction and the task's source evidence always outrank this skill'). This is a positive safety constraint that prevents the 'voice' persona from overriding primary agent instructions.
[INDIRECT_PROMPT_INJECTION]: The skill is intended to process external data, such as forum posts, Jira tickets, and pull requests, to generate responses. While this presents a theoretical attack surface, the skill itself possesses no dangerous capabilities (e.g., file writing, network exfiltration, or command execution) that could be exploited by an injection in the processed data.
[OBFUSCATION]: A thorough review of the text reveals no hidden Base64 strings, zero-width characters, or homoglyph-based URL spoofing. All URLs provided in examples (e.g., example.invalid) are clearly identified as placeholders.

voice-michael