Better prompts and specs for Claude Code — write them in TPEE
Claude Code is a CLI coding agent: it does what the prompt tells it. A vague brief gets vague edits; a structured spec gets a structured change. TPEE is an offline Windows editor for writing that spec first — then pasting the finished, formatted prompt into Claude Code. Looking for another target? See prompt templates for the Cursor IDE, prompt manager software for Windows or IDE & CLI command references.
By Jacques Botte, founder of Toptronic®. Last updated 12 September 2026.
How TPEE fits your Claude Code workflow
TPEE does not live inside the terminal — it runs beside it, offline. The workflow is simple:
- Draft the prompt or spec in TPEE's 12-section editor: persona, instructions, context, technical specifications, format, audience, tone, data, examples, graphics, MCP servers, agents. Each section shows a quality indicator and character count.
- Score it. TPEE rates the prompt 0–100 and tells you which sections are weak — the difference between "make it better" and a spec Claude Code can execute.
- Export with the Claude Code profile. TPEE supports 34 IDE and CLI export profiles; the Claude Code one arranges your sections into a formatted prompt ready to paste into the CLI.
- Reuse it. Save the prompt as a template, keep versions, and pick from 191 AI model profiles — the same structured brief works whether you delegate to Claude Code, Cursor or another agent.
The honest picture: TPEE is an authoring editor, not a Claude Code plugin. It won't edit code or run commands. What it does is make the prompt you paste into Claude Code structured, scored and repeatable — offline, in a one-time-purchase desktop app.
Why delegate with a written spec
| Typing it inline | Drafting it in TPEE |
|---|---|
| One-shot, retyped every session | Saved as a template, reused across sessions |
| Structure limited by the chat box | 12 explicit sections, including technical specifications and MCP servers |
| No feedback on prompt quality | 0–100 score with per-section feedback |
| History lost in scrollback | Prompt library and version history on disk |
Common questions
Is TPEE a plugin or extension inside Claude Code?
No. TPEE is a separate Windows desktop app (tpee.exe) that runs fully offline. You draft and score your prompts or specs in TPEE, then export a formatted prompt and paste it into the Claude Code CLI. There is no plugin to install and no network connection between the two.
Does TPEE support Claude Code as an export target?
Yes. TPEE ships 34 IDE and CLI export profiles, and Claude Code is one of them. The exported prompt follows the same 12-section structure — persona, instructions, context, technical specifications, format, audience, tone, data, examples, graphics, MCP servers, agents — arranged for a CLI coding agent.
Why write specs outside Claude Code instead of typing them inline?
Because a spec worth delegating is worth structuring. TPEE gives you a 12-section editor with per-section quality indicators, a 0–100 score, version history and reusable templates. You refine the spec once, score it, save it, and reuse it — instead of retyping a long brief every session.
Does TPEE send my code or prompts anywhere?
No. TPEE makes zero network calls at runtime — no account, no telemetry. You can verify it with a firewall. Your prompts stay in local JSON files on your machine; the only thing that leaves it is what you paste into Claude Code yourself.
What does TPEE cost?
TPEE is a one-time purchase (USD $20–$99 by plan, up to 12 months of updates included). There is no subscription, no account and no telemetry.
See for yourself
Compare the plans or watch the app in action. Also on this site: prompt templates for the Cursor IDE.