Well Made Decisions AI Skills: free AI agents for any assistant

Meet the agents

Each is a single set of instructions. Copy it straight into Claude or ChatGPT, or download the file, whichever you prefer. Grab the whole set with the button above.

How to use them

Each agent is a plain set of Markdown instructions you hand to an AI. Add it once (the Copy button on any card is the quickest way), then point it at your work by naming the skill and giving it a target: a website, a repo, or text you paste in.

Point a skill at a target:

/sentinel https://yoursite.comsweep a live site or web app for exposure
/cite https://yoursite.com/pricingcheck one page's citability and SEO
/frame https://yoursite.comcheck mobile layout and accessibility
Run Redline on this: then paste an email or copyproofread and voice-check any text

In Claude Code, type the slash command with the target. In Claude or ChatGPT on the web, just say it in plain language, for example “Run Sentinel on https://yoursite.com” or “Use Cite on this page: [paste the URL or text].”

Claude on the web or desktop app

Create a Project in Claude.
On the agent's card, click Copy (or expand View instructions and use Copy all).
Open the Project's custom instructions panel and paste. That's it.
Start a chat and point it at a target: say Run Sentinel on https://yoursite.com, or paste your text/code and name the skill (Redline this: [paste]). Answer the agent's setup questions the first time so it stays tuned to you.

Claude on the web can't edit your files directly, so the agent points to the exact fix and lets you apply it, instead of changing files itself.

ChatGPT (Custom GPT or Project)

Click Copy on the agent's card.
Create a Custom GPT (or a Project) and paste into the Instructions box.
Point it at a target: Run Cite on https://yoursite.com/pricing, or paste your text/code and name the skill. Answer its setup questions once.

A couple of lines in each agent are written for Claude Code (the slash command and the file auto-fixing). ChatGPT simply skips those and gives you its review and the fixes to apply yourself.

Claude Code (the developer CLI)

Unzip the download. Each agent is a folder with a SKILL.md inside.
Copy the folders into your project's .claude/skills/ folder (or into ~/.claude/skills/ to use them everywhere).
Run the slash command and point it at a target, for example /sentinel https://yoursite.com, /cite https://yoursite.com/pricing, or /sentinel on the repo you're in. Plain language works too ("run a security sweep on this repo").
On the first run the agent asks a few questions and saves a small profile in your project, so it fits your stack and standards. Every run after that starts straight away.

Claude Code can read and edit files, so these agents can safely fix the small stuff for you and flag the rest. This is the fullest experience.

Gemini, Copilot, Cursor, and other tools

These are just Markdown instructions, so they work anywhere you can give an AI a "system prompt" or custom instructions.
Gemini Gems / Microsoft Copilot: paste the copied text into the custom-instructions box.
Cursor, Windsurf, and coding agents: save the file as a rule, or paste it into the chat as context.
Anything else: paste the whole thing at the top of a fresh conversation, then hand over your content.

Any tool that can read and write files (like Claude Code) lets these agents fix the safe things for you. Chat-only tools advise instead. Either way you get the same review.

Tell us what you think

These tools are free and evolving. Your score and a way to reach you are all we need; the rest is optional but genuinely helpful.