Using the frameworks

This is the manual for operating the library well. It goes from “I can run one framework” to “I run, iterate, chain, and customize like a power user.” Read the tier you need and stop; the later tiers are here when you want them.

A framework here is not advice. It is a procedure an agent runs (or you run by hand) that ends in a concrete artifact: a ranked risk register, a weighted option matrix, an argument map. The artifact is the point.

---

Beginner

Anatomy of a framework page

Every framework page is layered so you can stop at any depth: a 20-second quick-facts card (the mechanism, the artifact, the evidence tier, when to use and when not), then the procedure and output format, then a worked example, then the full evidence dossier. You never have to read the whole page to use the framework.

Basic invocation

With an agent (Claude Code shown), call the skill by name with your situation:

/think-premortem "we're about to launch a free tier to drive signups"

By hand, open the framework page, copy its references/TEMPLATE.md, and follow the numbered procedure with a pen. The numbered steps are the method; the agent just makes them cheap to run and enforces the structure. See running by hand for the full workflow.

Describing your situation well (the SOFT check)

The framework supplies the structure. Your only job is to supply the situation. A good prompt usually has four things - remember SOFT:

• S - Situation: the real decision or problem, in one sentence, with the stakes.
• O - Options and constraints you already know (budget, deadline, what is ruled out).
• F - Facts versus unknowns: what you actually know, and what you are guessing.
• T - Target: what a good artifact would let you do next.

You do not need all four, and you do not need polish. A sparse prompt works because the framework, not the prompt, structures the output. The most common mistake is describing the framework instead of the situation - say what you are deciding, not how you think it should be analyzed.

Working with the artifact

The output is a document, not a chat reply. Save it. Put the risk register in the launch doc; paste the option matrix into the decision record. The value compounds when the artifact lives where the decision is made.

---

Intermediate

Iterating on an artifact

The first artifact is a draft. Push on it in plain language: “rank these by reversibility, not just impact,” “the second risk is the real one - expand its mitigation,” “add an owner column.” Specificity in the ask drives specificity in the artifact.

Chaining frameworks

The output of one framework is often the input to the next. Reframe a problem, then generate options against the new frame; compare options, then premortem the winner. When you copy an artifact into the next prompt, copy the part that matters (the chosen frame, the top three risks), not the whole thing.

Letting the advisor sequence for you

If you do not know which framework, or which order, describe the situation to the Framework Advisor. It returns a short Thinking Plan: the one or two frameworks worth running, in order, why each fits, and what to skip. It is the front door - use it whenever you are unsure, and overrule it when you have a reason.

Using a recipe instead of building a chain

When a chain recurs, it is a recipe: a fixed sequence shipped as one workflow. Stress-test a decision runs What-Would-Have-to-Be-True, then a comparison, then a premortem. Reach for a recipe when your situation matches its job; build your own chain when it does not.

---

Advanced

Power-user patterns

Named moves that combine frameworks for a recurring need:

• The Pre-Commit Gate. Before any one-way-door decision, compare the options, then run a premortem on the winner. Deciding and stress-testing are different jobs; do them in that order.
• The Steelman Swap. Before committing to the position you favor, argue the one you reject as well as you can (argument mapping or a red-team pass). If it survives, you are more sure; if it does not, you just saved yourself.
• The Evidence Check. Before you quote a number a framework produced, open its dossier. The library grades its own evidence; a “30% better” figure usually means something narrower than it sounds.
• The Frame-First Rule. When stuck, suspect the framing before the options. Restate the problem or audit its boundary before generating solutions to the wrong question.

Cross-referencing artifacts across one decision

A real decision produces several artifacts - a problem frame, an option matrix, a risk register. Keep them together and let them reference each other: the register stress-tests the option the matrix chose, against the frame the restatement set. The connected trail is worth more than any single artifact, and it is what the Showcase demonstrates end to end.

Integrating into your process

The frameworks earn their keep when they become habits, not one-offs. A decision journal before commitments and an after-action review afterward turn single runs into calibration over time.

---

Reference

By the job you need done

If you need to…	Start with	Or the recipe
Make a hard decision	Decision Option Review; Expected-Value Decision Tree	stress-test-decision
Reframe a stuck problem	Problem Restatement	reframe-problem
Generate real options	SCAMPER; Brainwriting; Morphological Analysis	expand-options
Stress-test a plan	Premortem; Red Team Light	stress-test-decision
Audit your reasoning	Evidence-vs-Inference Sort; Argument Mapping	audit-reasoning

Do not know which? The Framework Advisor picks for your specific situation. Or browse the six explore lenses (by job, by situation, by evidence, by artifact, the chooser, the map).

Next steps

You have the moves. Now: read the Prompt Gallery to see real prompts in three styles, watch a full decision in the Showcase, and check whether this actually works - the library measures its own routing and output quality, and publishes the numbers.