The Cash & Cache Library

Two slash commands — /eval-suite-wizard and /auto-research — that auto-generate a binary eval suite for any SKILL.md file, then run a self-improving loop: test → score → patch → retest until your skill hits a target pass rate or iteration cap.

The /build-my-pipeline wizard interviews you about your workflow, then auto-generates a complete chained multi-skill orchestrator — all stage skill files included — so you go from idea to running AI pipeline in one session.

A 4-prompt decision stress-testing framework: Steelman → Assumption Audit → Stress Test → Commitment Test. Run it before any product launch, pricing call, or strategic pivot. Available as a Markdown template and a one-click Claude Skill.

A Claude Cowork scheduled agent that runs every morning: searches the web, deduplicates against your Notion memory, applies an Intelligence Analyst skill to synthesise findings, and auto-delivers a tight briefing straight to Slack — no manual steps.

A custom Claude plugin with /evaluate and /compare commands that scores any AI tool across 8 capability dimensions and returns a calibrated workflow-fit verdict — structured, repeatable product research in seconds.

A drop-in Claude Skill (/prototype) for design-first builds — runs a 6-question design interview, generates the Claude Design master prompt, inspects the prototype against a 15-point PM checklist, turns tester feedback into a build/refine/kill verdict, and packages the validated design as a handoff brief for /spec.

A drop-in Claude Skill (/spec) that interviews your build idea like a product manager — encodes five PM frameworks, surfaces the hidden assumptions in your prompt, and produces a structured spec with user stories and acceptance criteria ready to push to Linear, Jira, or Notion via MCP.

A prospect research pipeline built entirely inside Google's ecosystem — Gemini Deep Research, NotebookLM, and Sheets. Two custom Gem commands produce a scored lead profile and outreach draft in under 20 minutes. Includes the 0-16 scoring rubric, copy-paste Gem instructions, all prompt templates, and a Google Sheet schema.

A Claude Skill that connects to your Notion database via MCP and reasons over it — mapping content gaps, spotting engagement patterns, and evaluating draft ideas against your own archive. Includes the full Skill file, reference file template, five prompt templates with gotchas, and a debugging guide for the five things that will break.

A PM's guide to directing a 5-agent BMAD team through a real product build — wired into Git and Jira so agents write commits, pull requests, and tickets without you touching either tool. Includes the Code Evaluator Skill for independent QA reviews before any change ships.

Full orchestrator file for the Breakthrough Method of Agile Development — 4-phase lifecycle (Analysis → Planning → Solutioning → Implementation), 7 specialist AI agents, quality gates, and slash-command workflows. Download and run in your own environment.

A two-phase visual branding system using ChatGPT Images 2.0 — build one reference image once, then generate covers, infographics, charts, and social carousels with one-line prompts forever. Includes the full Phase 1 setup prompt and per-format one-liners for every asset type you ship.

A 10-minute interview prompt that generates a personalised 5-prompt framework built around your work, audience, and specific frustrations — reusable across any AI tool without starting from scratch every session.

A structured Notion template that applies the CO-STAR prompting method inside Perplexity Computer — get sharper, more targeted outputs without inflating your token bill or burning through credits.

Step-by-step Notion guide for building a personal news aggregator that monitors any industry vertical and automatically surfaces relevant developments — structured workflow, no code required.

Exact workflow for producing a 2,000-word article, 5 Substack Notes, and a LinkedIn post in a single Cowork session — includes the folder structure, master prompt, and two-round editorial loop used to build a real Cash & Cache issue.

Three tested Claude-in-Excel workflows — dirty data cleanup, a 15,000-row Olympic dataset dashboard, and a 3-year SaaS P&L — with copy-paste prompts and an audit checklist for catching broken formula links before you present.

Notion guide for connecting Gemini to Gmail, Google Drive, and Google Calendar — so it can schedule meetings, triage emails, and organise files on plain-language command, without switching between apps.

Full walkthrough for a Gemini-powered research system that gathers intel, maps strategic plans, and produces visual infographics and interactive designs — from a single end-to-end workflow you can run repeatedly.

Practical guide to going from a rough idea to a polished visual using Claude Design — covers layout, colour, and asset generation end-to-end with no design background required.

Five Claude Code workflows — file manipulation, research automation, and life organisation — that require zero coding. Built for knowledge workers and operators, not engineers.

The five root-cause fixes from analysing over 40 Claude Skills failures — covering activation reliability, output consistency, and long-term maintenance so your skills keep working as your workflow evolves.

A week-long real test of Perplexity Comet across 4 newsletter workflows — research automation, content analysis, and more — with copy-paste prompt templates and an honest verdict on where it delivers vs. falls short.

A structured Notion audit template for identifying which AI tools you're actually using versus just paying for — includes a decision framework for cutting, consolidating, or committing to each tool in your stack.

Consultant-style divergence and convergence prompt sequence — go from wide ideation to a clear, defensible decision without losing creative range along the way. Structured for real strategic decisions, not brainstorming fluff.

A concise prompt that scopes and focuses any research task to your exact requirements — eliminates broad outputs and filler so every response is targeted, relevant, and immediately usable.

A multi-stage content framework guiding you through research, iteration, quality testing, and voice development — so your output sounds like you, not a generic AI template, and improves with every cycle.