20 Strategy Agents for Claude Fable 5.1: From Open Question to Executive Deck
Turn an open question into an evidence-backed decision and executive-ready deck with 20 bounded agents, one auditable work order, and three human approval gates.
AI Academy for Professionals
20 strategy agents turn Claude into a decision system
A clever prompt can produce an attractive answer. It cannot ensure that the right question was framed, evidence was sourced, real alternatives were compared, economics reconciled, or the recommendation survived challenge. Consequential work needs a decision work order, specialist agents, evidence rules, and human approval gates.
This free Oria package has two parts. The Strategy Agent Playbook explains how to prompt, route, control, and sequence the work. The companion ZIP contains 20 Claude Code agent definitions across Diagnose, Analyze, Decide, Execute, and Communicate.
Be the person your team asks about AI
Learn it once, then set the standard the rest of the team follows
Go to AcademyStart with a work order, not a persona
State who must choose what, by when, to maximize which objective, within which constraints. Then name the agent's bounded job, approved sources, exact output, evidence standard, and stopping point. For example: “Use Market Sizer to estimate the serviceable German market through 2030 using two independent methods. Show formulas, ranges, source dates, and conflicts. Do not recommend entry.”
Keep durable context in Project knowledge. Record recurring source, model-QC, interview-coding, or Oria slide-storyline methods as Skills. Agents own bounded workflows; Skills teach how your team repeats a method. If your team runs the same decision work in ChatGPT, the 50 GPT-6 Astra skills for strategy and finance carry the same six-stage backbone with the economics attached to every choice.
The 20-agent strategy system
Diagnose
Find the decision before searching for answers.
Problem Framer
Turns a vague topic into one owned, time-bound decision.
Situation Assessor
Builds a current-state baseline when leaders disagree on what is happening.
Hypothesis Builder
Creates falsifiable hypotheses and kill tests when confirmation bias is likely.
Issue Tree Architect
Decomposes the decision into prioritized, answerable branches.
Analyze
Build independent evidence streams, then reconcile them.
Market Sizer
Defines boundaries and triangulates market ranges for entry or growth choices.
Profit Pool Mapper
Shows where value is captured and how it is migrating across the chain.
Competitor War-Gamer
Grounds rival moves in evidence and simulates credible countermoves.
Customer Segmenter
Builds needs- and behavior-based segments from customer evidence.
Decide
Create real choices and make the economics visible.
Strategic Options Grid
Creates distinct choice systems, a reference case, and explicit trade-offs.
Pricing Strategist
Connects value, architecture, elasticity assumptions, and pricing pilots.
Business Case Builder
Models drivers, cash, returns, sensitivities, and break-even thresholds.
Portfolio Reviewer
Reallocates capital and attention across businesses, products, or initiatives.
Execute
Turn the approved choice into capabilities and control.
Operating Model Designer
Converts the choice into capabilities, governance, structure, and decision rights.
Initiative Prioritizer
Cuts the backlog to what capacity and dependencies can support.
Transformation Roadmapper
Creates phased delivery, accountable owners, milestones, and a 90-day plan.
KPI Architect
Links outcomes to leading indicators, thresholds, owners, and value realization.
Communicate
Make the approved recommendation survive the room.
Pyramid Story Builder
Builds one governing answer and an action-title narrative from approved analysis.
Decision Memo Writer
Records the recommendation, alternatives, conditions, dissent, and exact ask.
Hostile Q&A Simulator
Pressure-tests the call from executive and external stakeholder perspectives.
Executive Deck Packager
Produces an Oria-ready slide specification after the recommendation is approved.
Become Dangerously Productive with AI
For consultants, bankers and investors who want a compounding edge
Go to AcademyRoute the minimum team and keep humans at the gates
Do not invoke all 20 agents. A market-entry screen may need Problem Framer, Issue Tree Architect, Market Sizer, Customer Segmenter, Competitor War-Gamer, Business Case Builder, and Hostile Q&A Simulator. Parallelize independent evidence only; option design waits for evidence, execution waits for the choice, and deck packaging waits for the recommendation.
Humans approve three gates: the mandate and evidence plan, the option set or shortlist, and the recommendation before executive communication. Agents analyze and recommend. Accountable leaders decide.
Make every answer auditable
Label every material statement Fact, Estimate, Inference, Management assumption, or Unknown, with its source, date, and transformation. Triangulate high-impact estimates; expose formulas, units, and checks; preserve contradictions. Agents are read-only by default and cannot send, buy, upload, modify, publish, or act irreversibly without separate authorization.
Become Dangerously Productive with AI
For consultants, bankers and investors who want a compounding edge
Go to AcademyFrom approved recommendation to complex slides with Oria
Pyramid Story Builder creates the governing thought. Decision Memo Writer records the call and its conditions. Executive Deck Packager specifies the audience, action-title spine, exhibits, data lineage, appendix, and speaker cues. Claude structures the argument; Oria—AI for complex slides—turns the approved, traceable argument into polished executive slides.
Frequently asked questions
What is included in the Strategy Agent package?
The PDF explains the operating model, prompting rules, routing, evidence controls, and approval gates. The ZIP contains 20 Claude Code project-agent definitions, a quick start, a routing guide, and a reusable work-order template.
Are these Claude Skills or Claude agents?
They are Claude Code project agents, not Claude Skills. An agent owns a bounded workflow; a Skill records a reusable method. The Markdown agent bodies can also be used manually as specialist instructions in Claude.ai Projects.
How do I install the 20 strategy agents?
Unzip the pack at the root of your working folder so the files land in .claude/agents/. Add only approved context, complete the work-order template, and ask Claude to propose the minimum route before analysis begins.
Should I run all 20 agents on every strategy question?
No. Use the fewest agents that can materially improve the decision. Parallelize only independent evidence work, then keep option design, execution planning, challenge, and deck packaging in their dependency order.
Can I use the pack in Claude.ai Projects?
Yes, but not as native Claude Code subagents. Upload the approved context and work order to the Project, then use each agent file’s Markdown body manually in a chat and preserve the same human approval gates.
Where does Oria fit in the workflow?
Claude helps structure and challenge the strategy. After an accountable leader approves the recommendation, Executive Deck Packager creates the production brief and Oria turns the approved, traceable argument into polished complex slides.