Most agent failures trace back to one cause: the spec was underdense. This pack gives you the WOWHOW Spec-Density Scoring Sheet — a 6-axis rubric that scores any agent spec from 0–100 before you run it — plus 12 task-type templates pre-scored and ready to fill. It is for developers, solo builders, and AI product teams who write agent prompts today and debug them tomorrow.
What You Get
- Spec-Density Scorecard (CSV) — The 6-axis WOWHOW rubric: Trigger Clarity, Output Contract, Edge-Case Coverage, Tool-Permission Scope, State Handoff, and Rollback Safety. Each axis scores 0–20 with anchor descriptions at 0, 10, and 20. A composite score under 60 flags the spec as "Drift Risk".
- 12 Agent Spec Templates (Markdown) — One template per common task type: Data-Extraction Agent, Code-Review Agent, Email-Triage Agent, Document-Summarizer Agent, Scheduled-Report Agent, Web-Research Agent, Database-Migration Agent, Customer-Support Router, Content-Publishing Agent, API-Integration Agent, File-Organizer Agent, and Approval-Gate Agent. Every template ships with its own pre-filled Spec-Density score and annotated weak points.
- Scoring Worked Examples (Markdown) — Three real spec drafts scored live against the rubric: a low-density draft (score 34), a revised mid-density draft (score 71), and a hardened high-density version (score 94). Shows exactly which axis moved and why.
- README.md — Orientation guide, the theory behind Spec-Density, how to sequence the scorecard into a PR or prompt-review workflow, and a quick-reference axis cheat-sheet.
How to Use
- Open README.md first — read the axis definitions and the 60-point Drift Risk threshold.
- Pick the template closest to your task type from the 12 Markdown files.
- Fill every bracketed placeholder in the template. Unfilled brackets are automatic deductions on Trigger Clarity and Output Contract.
- Open the CSV scorecard. Score your completed spec against each of the 6 axes using the anchor descriptions. Enter your scores in column B.
- If your composite is below 60, the scorecard highlights the two lowest axes in red — fix those before running the agent.
- Re-score. A spec at 80 or above is cleared for production use according to the WOWHOW Spec-Density standard.
- Check the Worked Examples file to see how scoring decisions map to real draft language. The delta between the 34-score and 94-score versions is the fastest learning tool in the pack.
Who This Is For
- Developers writing Claude, GPT-4o, or Gemini agent prompts who want a pre-flight check before wiring tools or calling APIs.
- AI product leads who need a shared vocabulary to review agent specs in PRs without arguing about "good enough".
- Solo builders running automation workflows in n8n, Make, or custom Python who keep hitting unexpected agent behavior mid-run.
- Prompt engineers who already have instinct but want a repeatable, numberable standard they can report upward.
- Teams onboarding new contributors who need a spec format that enforces critical thinking rather than just filling fields.
Browse the full suite of developer productivity tools at WOWHOW Tools, or explore related AI-workflow resources in the WOWHOW product library. This pack is a one-time download — no subscription, no account required after purchase.
Claude