precon

Harvests idea discussions into fixed-format scope docs for downstream build planning.

Updated Jun 2, 2026
One-click install
npx skills add https://github.com/line7works/tony-skills --skill precon-line7works
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: precon
Source: https://github.com/line7works/tony-skills/tree/main/plugins/precon/skills/precon
Command: npx skills add https://github.com/line7works/tony-skills --skill precon-line7works

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Free-flowing idea discussions lose decisions the moment the conversation moves on, forcing teams to re-interview stakeholders before any build planning can start. This Skill captures every settled decision into a structured scope doc the moment it settles, so /blueprint can consume settled ground instead of re-asking questions. ## Core Features & Use Cases - Harvest-before-asking interview: Mines the existing conversation and repo docs first, then asks only numbered, batched frontier questions whose answers could plausibly differ from the given recommendations. - Tagged decision ledger: Records every line as decided, assumed, or parked (needs research, needs prototype, waiting on something), with an out-of-scope channel for rejected options, written to docs/scope/<date>-<idea>.md. - Cold-read exit test: Optionally summons fresh-context readers (Claude subagent or external models like GPT, Gemini, DeepSeek, Qwen) to report confusions in the scope doc, with verbatim output captured to a review doc. - Use Case: After a long chat fleshing out a new feature, invoke /precon to triage the idea (napkin, bounded, or architectural), run a few rounds of numbered questions, and end with a scope doc ready for /blueprint. ## Quick Start Type /precon after discussing an idea to harvest the conversation into a scope doc through a short round of numbered questions.

Frequently Asked Questions about precon

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I turn an idea discussion into a scope document?

Invoke /precon after discussing the idea. The skill harvests decisions already made in the conversation and repo docs, asks only remaining load-bearing questions in numbered rounds, and writes every settled decision to docs/scope/<date>-<idea>.md.

What is a scope doc for build planning?

A scope doc is a fixed-format record of an idea's intent, decisions, assumptions, parked items, and out-of-scope rulings. Downstream planning skills like /blueprint treat it as settled ground, so every line must trace to the user's words or an answered question.

Does precon do web research or look up external information?

No. The skill never leaves the property: no web access, research documents, or research subagents. Questions needing outside research are parked in the ledger tagged 'needs research' for the user to investigate between sittings.

Can precon send my scope doc to external AI models?

Only with your explicit word in that run naming a specific reader row such as gpt-astra, gemini, deepseek, or qwen. Nothing is sent externally by default, and the Claude cold-read row runs as a local zero-context subagent.

When should I skip writing a scope doc?

When triage classifies the idea as napkin-tier — describable in a sentence or two — the skill can recommend going straight to /blueprint. If you accept, no doc is written and the sitting ends with a read-back.

What happens when I re-invoke precon on an existing idea?

It reopens the existing scope doc found under docs/scope/, the older flat docs/ path, or ~/Documents, and continues the ledger in place. Parked and open items are revisited; a second doc is never forked.