CookiesForAI

discernment-nudge

Discernment Nudge

Discernment Nudge is Anthropic's agent skill that ends a substantive Claude answer with two or three specific follow-up questions, so you can check its facts, reasoning and assumptions.

Agent skill by Anthropic Official · Updated

Install Discernment Nudge

In Claude Code, add Anthropic's skills marketplace

Terminal
/plugin marketplace add anthropics/skills

Then install the discernment-nudge plugin

Terminal
/plugin install discernment-nudge@anthropic-agent-skills

Or install only this skill, for any agent, with the skills CLI

Terminal
npx skills add anthropics/skills --skill discernment-nudge

Or paste this into your coding agent: Install the agent skill discernment-nudge from github.com/anthropics/skills. A skill can include scripts that run on your computer, so read its source first.

What Discernment Nudge does

Discernment Nudge is an Anthropic skill built on three habits from the AI Fluency framework: checking facts, questioning reasoning and noticing missing context. It loads after Claude gives an answer you may act on: advice or a recommendation, a drafted plan, pitch, proposal or email, estimates or projections, an interpretation of data, factual claims you may rely on, or a multi-step argument.

With Discernment Nudge, Claude answers in full first. It then adds the line "A few things worth a second look:" followed by two or three short questions you could send back. Each question points at something specific in the answer, such as a figure, a step in the reasoning or an assumption about your situation, and stays under about 120 characters. Nothing follows the list.

The Discernment Nudge appears at most once per conversation. Claude leaves it off for creative writing, casual chat, code you will run, simple lookups and purely educational explanations. It also stays silent when you already asked it to verify or cite, asked for the quick version, asked it to review your own work, gave it your own material to summarize, or asked for its opinion.

When to use Discernment Nudge

  • You use Claude for business, money, health or career advice and want a prompt to check it before you act.
  • You get cost or timeline estimates from Claude and want to know which numbers to verify.
  • You want to build the habit of questioning AI answers without a warning on every reply.

When to pick something else

  • You want sources checked on every reply: the nudge appears at most once per conversation and skips code, lookups and creative writing.

Which agents Discernment Nudge works in

Anthropic documents Discernment Nudge for Claude Code, Claude API. The open skills CLI also installs it into 78 agents, including Claude Code, Codex, Cursor, Gemini CLI, GitHub Copilot, OpenCode (we listed it with the CLI on October 1, 2026). See where each agent looks for skills.

Anthropic wrote Discernment Nudge for Claude's answers, from its AI Fluency course, but the instructions use no Claude-only tool.

Discernment Nudge license

Discernment Nudge is published under Apache-2.0. Apache-2.0 comes from the LICENSE.txt in the skill's own folder (read 2026-10-01); the anthropics/skills repository has no root license file. In practice you may use, change and share the skill in any agent, commercially too, as long as you keep the license text and notices and mark your changes.

  • Academy Guide: Academy Guide is Anthropic's agent skill that ends Claude's answers about using Claude with one or two matching Claude Academy courses or tutorials, and only when the match is strong. (179,326 repository stars on October 2, 2026)

More from Anthropic

  • Algorithmic Art: Algorithmic Art is Anthropic's open-source agent skill for generative art: the agent writes a short art philosophy, then builds an interactive p5.js piece with seeds and parameter sliders. (179,326 repository stars on October 2, 2026)
  • Brand Guidelines: Brand Guidelines is Anthropic's open-source agent skill that holds Anthropic's own brand colors and fonts, so an agent can give a document, slide deck or web artifact the Anthropic look. (179,326 repository stars on October 2, 2026)
  • Canvas Design: Canvas Design is Anthropic's open-source agent skill for static visual work: the agent writes a design philosophy, then makes a poster or art piece as a finished PNG or PDF file. (179,326 repository stars on October 2, 2026)
  • Claude API: Claude API is Anthropic's open-source agent skill for code that calls Claude: current model IDs and prices, guides for seven SDK languages plus cURL, and guided migrations to newer models. (179,326 repository stars on October 2, 2026)

Lists that include discernment-nudge