How to use the Consciousness Council agent skill

What an agent skill is and what this one does

An agent skill is a folder of instructions that an AI agent loads when a request matches the skill's description. This skill is defined in a SKILL.md file. It tells the agent to run a three-phase exercise: summon a council of 4 to 6 members from 12 thinking archetypes, have each member give a position with its reasoning and risks, then write a synthesis.

Before deliberating, the agent builds an evidence brief that separates given facts, externally verified facts with sources, assumptions, unknowns and value judgments. The output is a conditional recommendation, open questions and a next step.

• The Strategist looks at competitive moves and second- and third-order effects.
• The Empiricist asks what the evidence actually shows.
• The Ethicist asks who benefits and who is harmed.
• The Contrarian tests the strongest assumption behind the leading option.
• The viewpoints are generated by the same AI system. They are not independent experts or independent evidence.

Install the consciousness-council skill

Download the ZIP of the skill from this page. Then choose one of two routes.

For Claude Code, unzip it into ~/.claude/skills/ to make it available for you across projects, or into a project's .claude/skills/ folder to make it available only in that project.

In the Claude apps, upload the skill under Settings → Capabilities → Skills.

Trigger the skill with natural requests

You do not need a special command. The skill's description lists the kinds of requests it matches: asking for a council, a panel, a devil's advocate analysis, a "mind council", or a deliberate comparison of perspectives on a difficult choice.

Say what the decision is and what you are weighing. Add constraints and your objective if you have them. The agent will state any material missing information as an assumption or a question.

• Convene a council to decide whether to take the job offer or stay with my current team.
• Run a devil's advocate analysis on my plan to launch the product in March.
• Use a mind council to compare two study designs for my experiment.
• Give me a panel view on whether to raise prices, including the ethical side.

What you need and what it does on your machine

The source lists no required dependencies. The skill does not run scripts and does not use the network. It runs within the current conversation and has no bundled code, service API, credentials or external connection.

For facts that could change the recommendation, the skill directs the agent to use the host's available research tools for verification. If verification is not available, the conclusion stays conditional.

• Requires: nothing listed.
• Runs scripts: no.
• Uses network: no.
• Licence: MIT.

Read the output with the right level of trust

The council is a decision aid, not proof. Agreement among simulated viewpoints does not establish that a conclusion is valid, and the number of perspectives does not add accuracy. A simulated Empiricist is not a source.

Check any consequential factual claim against original sources yourself, and treat the confidence level as confidence in the proposed next step, not in the eventual outcome.

• Verify key facts with original sources or your own data.
• Do not use the output as a substitute for a doctor, lawyer or financial adviser.
• Expect the synthesis to show open questions, not only a single answer.

Tips for better councils

Give the agent specific options, including postponing the decision if that is realistic. Supply facts you already know so they are treated as given facts rather than guesses.

The skill also includes advanced configurations in references/advanced-configurations.md, such as a startup decision council or a technical architecture council. Use these when your question fits one of those domains.

• List at least two concrete options.
• State your objective in one sentence.
• Ask for a specific council size if you have one in mind: quick (3 perspectives), standard (4 to 6) or deep (6).