The Agent Skills Directory

[SAFE]: The skill serves as a structured library and methodology for prompt engineering. It does not contain malicious code, credential harvesting, or unauthorized data access patterns. All external links point to official developer documentation.\n- [COMMAND_EXECUTION]: Several levels (2, 3, 4, 6, 7) provide examples of using tools like Bash, Task, and Write. These are presented in the context of educational templates for automating developer workflows, such as starting a server, running git commands, or managing prompt files. The 'Self-Improving' level (Level 7) specifically includes logic to update its own 'Expertise' section with discovered patterns, which is a documented feature for knowledge accumulation.\n- [EXTERNAL_DOWNLOADS]: The documentation references official Anthropic URLs for tool and command reference. These are trusted informational sources and do not represent a remote code execution risk.\n- [PROMPT_INJECTION]: The skill provides templates that ingest user requirements to generate prompts. While this creates a potential surface for indirect prompt injection, the skill's methodology emphasizes structured sections and clear workflows, which are defensive practices against unintended instruction obedience.

writing-prompts