Prompt library

Battle-tested prompts for diagnosis, framework generation, writing, and orchestration. Each one ships with inputs, models tested on, and worked examples.

9 prompts

discovery

Discovery Call Skeptic

Stress-test a discovery call summary by surfacing the things the prospect didn't say, the misalignments hiding in agreement, and the configuration risk you're about to walk into.

claude-sonnet-4-6 · gpt-5

diagnostic

The Week Test

Audit last week's calendar and output to surface where your judgment was actually load-bearing versus where you were doing work a model could have done.

claude-sonnet-4-6 · claude-opus-4-7 · gpt-5

diagnostic

Before you pick a tool, figure out what layer of the build stack you actually want to live on. This diagnostic walks you through your last three projects and tells you whether you are building above or below your real capacity.

claude-sonnet-4-6 · claude-opus-4-7

framework

Cohort Insight Engine

Surface the non-obvious pattern across a cohort of client cases — the configuration that keeps repeating that nobody has named yet.

claude-opus-4-7

framework

The Canon Extractor

Surface the few things you actually believe are true about your audience and your domain, the load-bearing beliefs that anchor everything else you say. Output is the central canon that should sit underneath your content system.

claude-opus-4-7

framework

Process Prompt Builder

Stop writing prompts directly. Have the AI interview you about how you actually do a recurring piece of work, then synthesize a reusable process prompt that interviews the next user, plans, and executes.

claude-opus-4-7 · gpt-5

writing

Residue Capture

At the end of a piece of real work, capture the things that did not make it into the deliverable. The asides, the dead ends, the moments of judgment. This is the substrate the AI needs to later produce work that sounds like you.

claude-sonnet-4-6 · claude-opus-4-7

orchestration

Structured Object as Agent Memory

Before you build a database or a vector store for your agents, audit the structured objects you already use. A Notion page, a Trello card, a GitHub issue, a CRM record can all serve as agent memory if their fields are used with discipline.

claude-opus-4-7

audit

Skeptic on the Build

Before you commit a week to building an AI system, run the build proposal past a skeptical advisor that has no investment in seeing it shipped. Surfaces the cases where you are about to over-engineer, under-scope, or solve the wrong problem.

claude-opus-4-7 · gpt-5

Prompt library

discovery

Discovery Call Skeptic

diagnostic

The Week Test

Abstraction Layer Diagnostic

framework

Cohort Insight Engine

The Canon Extractor

Process Prompt Builder

writing

Residue Capture

orchestration

Structured Object as Agent Memory

audit

Skeptic on the Build