responsible-ai-evaluator skill - #10
Conversation
|
|
||
| The evaluation is based on Sparkgeo's 10 Principles for Responsible AI Use: | ||
|
|
||
| 1. **You own the output** — You are fully responsible for AI-generated content |
There was a problem hiding this comment.
How does this work if it's not a question ?
There was a problem hiding this comment.
It prompts you to ask a question and describe the task you want to evaluate for responsible ai use.
There was a problem hiding this comment.
I misunderstood your comment, and i may still not understand it, but the principles don't have to be in question form for the evaluator to ask about them. The evaluator creates its own questions to evaluate whether your idea fits with the principles.
|
|
||
| ## When to Offer Guidance | ||
|
|
||
| The user can ask "What does principle X mean?" or "How do I satisfy principle Y?" at any time. Provide clear, actionable guidance using the principle's description and guidance notes. |
There was a problem hiding this comment.
Some principles are more statements then questions ?
There was a problem hiding this comment.
I don't really see that as a problem. For example, "you own the output". The evaluator attempts to determine if you will truly own the output.
2 participants
Add Responsible AI Evaluator skill
Adds an interactive Claude Code skill for evaluating whether a task is suitable for AI based on Sparkgeo's 10 Principles for Responsible AI Use.
How It Works
User describes a task → Claude asks several clarifying questions → Shows final scorecard and recommendation (Suitable / Proceed with Caution / Not Suitable).
Usage
Copy
SKILL.mdto.claude/rules/or paste into Claude context, then ask Claude to evaluate a task using the evaluator. Works in Claude Code, Cursor, and any Claude interface.Files
skills/responsible-ai-evaluator/SKILL.md— Evaluation instructionsskills/responsible-ai-evaluator/README.md— Usage guideNote: Results are AI-generated. Users must apply their own judgment.