Skip to main content

Documentation Index

Fetch the complete documentation index at: https://mintlify.com/8BitTacoSupreme/flowstate/llms.txt

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

FlowState is a CLI-first orchestrator that prepares your project for agentic development. Run one command to kick off a five-step pipeline: generate structured context files, run targeted LLM research, pressure-test your strategy, scaffold GSD project management artifacts, and audit your repo health — all from your terminal.

Introduction

Learn what FlowState is, how the pipeline works, and when to use it.

Quickstart

Install FlowState and run your first pipeline in under five minutes.

CLI Reference

Every command, flag, and option documented with real terminal output.

Concepts

Understand the pipeline, persistent memory, and the Claude bridge.

​
How It Works

FlowState runs a five-step pipeline in sequence, persisting state to flowstate.json after every step so it can resume from failures.
1

Answer the intake interview

FlowState prompts you for your research focus, core problem, 10x vision, milestones, and architecture pattern. Answers are stored in flowstate.json and reused on subsequent runs.
2

Generate context files

Five deterministic files are written in under a second: PROJECT.md, ROADMAP.md, config.json, CLAUDE.md, and research/brief.md. No LLM required.
3

Run research and strategy

FlowState splits your research focus into topics and calls claude --print once per topic (~30 seconds each), then runs a single strategic pressure-test call (~75 seconds). Findings are stored in a local SQLite memory store and injected into future runs.
4

Launch GSD and discipline phases

Use flowstate launch gsd 1 to get the exact Claude Code command for running GSD phases natively. The discipline audit checks your git repo, test config, and hooks in pure Python — no LLM needed.

​
Key Features

Persistent Memory

Research findings and strategy decisions accumulate in a SQLite FTS5 store. Every new pipeline run benefits from prior knowledge.

Claude Bridge

A thin wrapper around claude --print with system prompts, tool permissions, model selection, and budget controls.

Dry-Run Mode

Test the full pipeline without calling the Claude CLI. Every adapter has a mock mode that writes realistic placeholder artifacts.

Flox Environment

One flox activate gives you Python, Node, Claude Code CLI, and MCP servers — fully reproducible across machines.

Build docs developers (and LLMs) love

Get started for freeTalk to us