Hire an AI assumptions and evidence tracker

Everything it can hand back

Your task sets the actual outputs. The specialist picks the methods that cover them.

• A prediction, a sinking fact, a bet and a date for each belief
• Every fact for and against, logged with its source
• A keep, change or stop call for each belief that moved
• A plain flag for any belief still held against strong evidence
• The hardest questions on a quiet belief, and a learn, borrow, start small or pass plan
• What the record cannot show, and the fact that would settle it

How it does the job well

• Pin down the exact question, scope and the standard of evidence needed before searching.
• Prefer primary sources, and say how strong each source is.
• Separate what is established, debated and unknown, with confidence levels.
• Cite specifically: author, work, date or page where possible.

Mistakes it avoids

• Logging the facts for a belief first.
• Treating a belief with no evidence against as settled.
• Extending a bet's date to avoid calling it failed.
• Keeping a belief because the plan depends on it.
• Presenting one source or a popular claim as settled.
• Inventing citations, quotes or page numbers.

What a good result looks like

An example, not your task: Your plan rests on beliefs about buyers and prices that nobody has tested in a year.

• This month. One line with beliefs tracked, how many moved, how many are quiet and how many bets are due.
• The register. Each belief with facts for and against, the verdict and the next check date.
• Moved beliefs. New facts against and for, a keep, change or stop call, and what to learn next.
• The record says otherwise. Any belief you still hold against strong evidence, with your stated reason.
• Quiet beliefs. Hardest questions, how many you can answer, and a learn, borrow, start small or pass plan.
• Bets. Each bet that is due, with what was predicted, what happened and a held, failed or unclear verdict.

How the work is checked

• Every belief has a written prediction and the fact that would sink it.
• Facts against were logged before facts for, each with a source and date.
• Every verdict rests on the log, not on the person's preference.
• No bet was marked held or failed without its numbers.
• States which checks passed, failed or could not run, and names missing inputs.

These checks have not been run against a model yet. Your AI uses them to check its own work.

A second opinion

Before it hands back, your AI reviews the draft the way Charlie Munger and Charles Darwin would. This is a way of reviewing the work. It does not mean they wrote or endorse it.

Capabilities

• Set up each new belief. Writes what it predicts, the fact that would sink it, a small bet and a check date.
• Log the evidence. Every fact against first, then the facts for, each with a source and date.
• Review the moved ones. For each belief with new evidence, a keep, change or stop call and the next thing to learn.
• Check the quiet ones. After three months with no evidence, asks the hardest follow-up questions and marks the ones you cannot answer.
• Close the gap when found. Learn it, borrow help, start small or pass, with a first-week plan.
• Check the bets. On each check date, reads the bet against the prediction and writes held, failed or unclear with the numbers.
• Say what the record cannot show. One line per belief on the limit of the evidence and the fact that would settle it.
• Log and hand back. Updates the register and stops. You change the plan.

How the role works

1. Checks your task, context and available tools.
2. Picks the methods needed and prepares a work plan.
3. Carries out supported tasks, with each method's limits and review gates.
4. Checks the real deliverables and corrects supported faults.
5. Hands back files, evidence and unresolved gaps.

Your specialist's working setup

Download the base specialist without a task form or account. README.md and SETUP.md explain how to use the folder. SOUL.md defines its scope. INTAKE.md asks missing questions inside your AI. PLAYBOOK.md, SKILLS.md, BEST_PRACTICES.md, CHECKS.md and HANDOFF.md guide the work and its checked deliverables.

Complete associated skills, supporting files and licences are included. Choose only the methods needed for the actual task. Keep context, working files and accepted feedback in your AI project.

Use this specialist through MCP or the API

Call prepare_specialist with role_id monthly-belief-evidence-keeper only. No task is required to load it. Your AI asks what job you need done, what you want back and which relevant context it can use.

For task-led hiring, call find_specialist, explain the suggested specialists and let the user accept or change the team.

Complete ZIP: Download base specialist. API: GET /api/v1/specialists/monthly-belief-evidence-keeper/pack. The host AI supplies its own model, tools and execution; a pack does not connect accounts or run a task.

Inputs and helpful context

The specialist starts from the task brief. Only active methods require their inputs.

For the primary method, when used

• Five to ten beliefs your plan rests on, each in one sentence
• This month's evidence, with where each fact came from (advised)
• Your plan, so a changed belief can be traced to what it holds up (optional)
• The register, once it exists

Helpful context (optional)

• Customer interview notes from the month
• Numbers that bear on a belief, such as cancel reasons or ad results by angle

Tools and access

• Claude Code: It can keep one file per belief and the register between months. You start the run on the first Friday of each month. It does not run on a timer unless you set up a scheduled task or connect a scheduler.
• Any good chat AI: Paste the agent, your beliefs and the month's evidence. Keep the register in a note yourself.
• Any good chat AI: Turning a belief into checkable claims is text work that any strong chat AI does with this skill.
• A shared note or doc: Lets others see the bet and the date, which keeps you honest.
• Any good chat AI: Setting up the log and running the review is text work any strong chat AI handles well.
• A spreadsheet or notes app: A table you open every day keeps the log alive better than a chat that scrolls away.
• Any good chat AI: Weighing notes against beliefs is text work any strong chat AI does well with this skill.
• A shared document: Keeping beliefs, notes and calls in one place lets a partner check your reasoning.
• Any good chat AI: Writing the questions and marking answers is text work. Claude, ChatGPT and Gemini all do it with this skill. It needs no files.
• A notes app or document: Keeps your answers and the result in one place you can come back to.
• Any good chat AI: This is text reasoning that any strong chat AI handles with the skill.

Functions and deliverables

• Plan assumptions and evidence tracker: Track evidence for and against your plan’s key assumptions. Review what changed and choose what to test next.
  • A prediction, a sinking fact, a bet and a date for each belief
  • Every fact for and against, logged with its source
  • A keep, change or stop call for each belief that moved
  • A plain flag for any belief still held against strong evidence
  • The hardest questions on a quiet belief, and a learn, borrow, start small or pass plan
  • What the record cannot show, and the fact that would settle it
• Prediction and bet planner: Turn a belief into a checkable bet. Get what it predicts, what would prove you wrong, a small stake and a date to check.
  • The belief in its simplest form
  • 2 to 4 things it predicts that you could check, each with where to look and by when
  • The result that would prove you wrong, written before the test
  • A small stake, who holds it and a date to check
  • Rival explanations set out fairly, with the one you prefer
  • Blanks to fill in on the date: result, won or lost, what changed your mind
• Assumption evidence tracker: Keep a log of facts that go against your plan or theory, then run a weekly review. Built to track contrary evidence before you commit.
  • A one-line version of your belief
  • A log table ready to fill in, with the rows you already have
  • A weekly review script and a restated belief after each review
• Idea evidence reviewer: Check your customer talk notes against what you believed. Get a keep going, change or stop call on your idea, plus your next three talks.
  • Each belief marked supported, weakened, untested or wrong, with the evidence
  • The traps that may be skewing what you heard
  • A keep going, change or stop call, with the reason and a stop rule
  • Your next three questions and the next three conversations to have
• Subject knowledge tester: Test how well you really know a topic. Get six hard expert questions, each answer marked real or only talk, and the gaps to close.
  • Six hard questions an expert would ask
  • Each answer marked real, partial or only talk, with where the knowledge came from
  • A score out of six and a one-line verdict on your claim
  • The places where you sounded surer than you were
  • Gap questions you can look up or ask an adviser, ranked by how much they could change your decision
• Knowledge gap decision planner: Not sure whether to learn it, borrow help or pass? Paste your knowledge gaps and get a clear choice with a first-week plan and a test date.
  • A choice: learn, borrow, start small or pass, with reasons
  • The few big ideas to learn first, if learning
  • The checks for any helper you borrow
  • A first-week plan and a date to test yourself again
• Research paper and code note writer: Share a paper, repo or messy notes. Get claims, methods and evidence sorted into a clear research folder.
  • A research folder with claims, methods and results
  • Every claim linked to its supporting table or log
  • Results tables copied with exact numbers
  • A record of dead ends and decisions
  • A short report on what was filled in and what was missing
  Source: Orchestra Research.
• Charlie Munger: Stress-test a plan using Charlie Munger’s recorded approach. Turn it upside down, examine incentives and ask where your knowledge ends.
  • Your problem worked on the way Charlie Munger would: their questions, their method, their push-back
  • The answer or the work, in Charlie Munger's voice, using their ideas by name
  • Anything that goes beyond their public record, marked
  • One suggested next step
• Charles Darwin: Build a case using Charles Darwin’s recorded approach. Keep facts that challenge your idea, face the hardest objections and show where the evidence stops.
  • Your problem worked on the way Charles Darwin would: their questions, their method, their push-back
  • The answer or the work, in Charles Darwin's voice, using their ideas by name
  • Anything that goes beyond their public record, marked
  • One suggested next step

Tasks you can use it for

• Your plan rests on beliefs about buyers and prices that nobody has tested in a year.
• You keep remembering the evidence for your idea and losing track of the evidence against.
• You want a monthly habit that tells you when a belief has quietly died.

Limits to check before starting

• You need to weigh one single decision once: paste it into your AI as a plain question, or ask a trusted person to challenge it.
• You want the plan changed for you: this agent never edits the plan, so make that call yourself or with your team.
• You have no beliefs written down yet and no evidence to check: first write five to ten one-sentence beliefs, then come back.

Best way to use this role

Give it one clear outcome, the actual files and a definition of done. Keep the work in one AI project so it can reuse checked context and earlier files.

• Start a chat with the agent. In Claude Code, add it as a skill or save it in .claude/agents, then give it your beliefs.
• Write each belief in one sentence. On the first run, approve the prediction and the sinking fact for each.
• Each month, paste the new evidence with its source. Include the facts that hurt.
• Read the moved beliefs and the flags. Change the plan yourself where the record says to.
• Run it again on the first Friday of next month, and on any bet's check date.

Read it as data

Read the complete role setup as JSON. Connect your AI to the AI Blocks MCP library and read aiblocks://roles/monthly-belief-evidence-keeper for its role instructions and pinned sources.

This is a prepared AI role. No task has run. Your AI carries out authorized work using its available tools and reports the actual deliverables and checks.