The Agent Skills Directory

[SAFE]: The skill contains no executable code or malicious instructions. It is purely informational, providing guidelines for agent behavior during software development tasks.
[EXTERNAL_DOWNLOADS]: The skill references established, highly trusted sources such as RFC 2119, Google's SRE book, Martin Fowler's blog, and NASA's product validation guidelines. These references are used for educational purposes and do not involve remote code execution or untrusted downloads.
[COMMAND_EXECUTION]: While the skill mentions running project commands (such as formatting, linting, and npm install), it does not provide specific hardcoded commands to execute. Instead, it instructs the agent to consult the project's own documentation to find the correct, intended invocation for that specific environment.
[DATA_EXPOSURE_EXFILTRATION]: The skill mentions sensitive configuration and discovery files but only in the context of documenting project operations or following vendor guidelines. No instructions are present to read or transmit sensitive data like credentials or SSH keys.
[INDIRECT_PROMPT_INJECTION]: As a development-oriented skill, it necessarily interacts with external data (project source code and documentation). The skill mitigates risks by requiring the agent to prioritize the project's documented intent and seek human approval for undocumented operations.

software-development