The Agent Skills Directory

[SAFE]: The skill consists entirely of natural language instructions (SKILL.md) and evaluation metadata (evals/). There are no scripts, binaries, or active code components that could execute on a host system. While there are Python fixtures in the evals/ directory, these are static examples used for testing the agent's logic and are not executed by the skill itself at runtime.
[SAFE]: The skill explicitly defines professional and ethical boundaries. It instructs the agent to never infer or diagnose conditions, maintain medical/safety caveats, and prioritize accuracy and consent over brevity in high-stakes situations.
[SAFE]: No remote code execution patterns, external downloads, or network operations were detected. All external references in THIRD_PARTY_NOTICES.md point to well-known repositories (GitHub) for licensing and attribution purposes.
[INDIRECT_PROMPT_INJECTION]: The skill processes user-supplied queries to adapt formatting. However, because it lacks privileges to write files (outside of standard agent permissions), execute shell commands, or perform network requests, the risk of an indirect injection leading to data exfiltration or system compromise is negligible.
Ingestion points: User prompts (SKILL.md)
Boundary markers: Present (instructions explicitly command the agent NOT to mention ADHD unless the user does and to stick to factual/safety requirements)
Capability inventory: None (no tools or commands requested)
Sanitization: N/A (standard LLM guardrails apply)

adhd-friendly