Perfect a rough request or draft prompt using relevant context, attachments, files, and the minimum necessary skills or tools, then answer or execute it. Use when the user invokes $prompt or asks to improve, rewrite, sharpen, expand, contextualize, or optimize their prompt before acting. For code-changing prompts, verify task ownership and use an isolated checkout when needed; preserve verified same-task continuation. Do not use for ordinary code or content review unless prompt improvement is requested.
86%
Does it follow best practices?
Run evals on this skill
Adds up to 20 points to the overall score
Low
Low-risk findings worth noting
The skill exposes the agent to untrusted, user-generated content from public third-party sources, creating a risk of indirect prompt injection. This includes browsing arbitrary URLs, reading social media posts or forum comments, and analyzing content from unknown websites.
The required workflow only ingests outsider-authored text when the agent actively fetches web pages for current or source-dependent facts (SKILL.md step 7 and the browsing guidance), so the attacker can publish poison but cannot force delivery; no monitored inbox, feed, or issue queue is read.
BigSimmo/PsychSift
fada614
If you maintain this skill, you can claim it as your own. Once claimed, you can manage eval scenarios, bundle related skills, attach documentation or rules, and ensure cross-agent compatibility.