Skip to main content

Documentation Index

Fetch the complete documentation index at: https://mintlify.com/XxYouDeaDPunKxX/GPT-PF-Chat-GPT-Project-Forge/llms.txt

Use this file to discover all available pages before exploring further.

A guided ChatGPT session lets you use the AI to help close scope, classify sources, and compile opening artifacts before the real project ever starts. The Project Forge workspace acts as a controlled preparation room — a place where ambiguity is resolved and structure is built — so that the target project can open from explicit files instead of reconstructed chat memory.

​
The rhythm

1

Keep the Forge workspace clean

Open a new chat or use a clearly isolated project for Project Forge preparation. Do not mix preparation work with the target project itself. If the environment allows persistent memory or chat history reference, be deliberate about disabling or isolating it. Contamination from prior sessions can make the preparation look coherent while actually pulling from unintended context.
2

Provide the Forge files

Give the AI AI_START.md and the four canonical files in read order: 00_SCOPE.md, 01_RULES.md, 03_ARTIFACTS.md, 02_PROTOCOL.md. These are the governing surfaces. The AI reads AI_START.md as startup routing, then reads the canonicals for authority. Do not skip the canonicals — without them, the AI is operating on protocol alone, which can be misread as permission.
3

Describe the target project

Tell the AI what target project is being prepared. Provide the objective, scope constraints, relevant source material, and expected outputs in plain language. Be explicit about what is in scope and what is out. The AI will help you close what is ambiguous, but the operator controls what counts as real input.
4

Let ChatGPT help compile the opening package

Allow the AI to work through the standard flow: close the object, close the target, classify information, validate sources, assess readiness, and emit only the authorized artifacts. The AI should surface gaps, flag missing fields, and stop when ambiguity changes the move rather than filling silently past it.
5

Review the package

Before using the package, review each artifact as the operator. Check that the brief reflects the actual objective and scope. Check that the SSOT contains only validated, approved material. Verify that any handoff or transfer artifact was genuinely needed. The AI builds the draft; the operator confirms what is real.
6

Open the real target project from those files

Create or open the target project and start it from the opening package. The target project reads the entry point, brief, and initial SSOT as its starting ground. It does not need access to the preparation chat. If the package is complete, the target project can begin without the preparation session as hidden state.

​
What to give the AI

A good startup package for the guided session contains:
  • AI_START.md
  • The four canonical files (00_SCOPE.md, 01_RULES.md, 03_ARTIFACTS.md, 02_PROTOCOL.md)
  • Only the artifacts that actually exist for this case
  • Only the source pack you want the AI to use
Do not give the AI old chats as if they were authoritative. Do not pass in hidden assumptions, mixed notes that have not been classified, or every available source “just in case.” Volume creates noise before it creates clarity. Material that has not been explicitly validated should not silently enter the preparation as if it were a reliable basis.

​
The preparation room concept

Project Forge separates the work of preparation from the work of execution.
  • The Project Forge workspace is the preparation room — where scope, sources, and structure are resolved before the project begins.
  • The target project is the destination — the real ChatGPT project where the work happens.
  • The opening package is the bridge — the explicit files that allow the target project to start from a known, controlled basis instead of chat reconstruction.
This separation is the core of what Project Forge does. The preparation room can be messy and iterative. The opening package it produces must be clean and explicit. The target project should never depend on the preparation room as hidden state.

​
Three session shapes

Do not mix these three modes casually. If you do, you usually create confusion about what is real state and what is only a test.
Use this shape when you are opening a new target project from scratch and no prior artifacts exist.Typical inputs:
  • AI_START.md
  • The four canonicals
  • Explicit source materials relevant to the target project
What to do:
  • Close the target and scope
  • Validate sources
  • Instantiate the brief and initial SSOT
  • Assess readiness
  • Open the target project from the resulting package

​
Minimal operator instruction

The shortest correct way to start a guided session is: 
Read AI_START.md, then use the Project Forge canonicals to prepare the opening package for this target project.
Follow this with the target description, source material, constraints, and expected outputs. The AI will use the canonicals to govern the preparation and emit only the artifacts the case actually requires.

Build docs developers (and LLMs) love

Get started for freeTalk to us