FlightPathAI workflow automation for VS Code and the command line

Run structured multi-step AI workflows, validate the result with scripts, and keep working across harnesses without rewriting your prompts.

Quick start

Workflow schema

GitHub

One workflow format

Write workflows in YAML, JSON, or simple Markdown prompts. The same schema works across the extension and the standalone CLI.

Seven harnesses

Switch between Copilot CLI, Claude CLI, Codex CLI, OpenCode CLI, Qwen CLI, Gemini CLI, and Copilot Chat.

Validation and retries

Gate prompts with end conditions, pre-conditions, and script steps so workflows can stop, continue, or retry based on real project output.

Stateful execution

Resume sessions across steps, inject prior context, and let final steps extend the queue through the built-in FlightPath MCP server.

What FlightPath is for

FlightPath turns AI assistants into a repeatable workflow runner. Instead of pasting one prompt at a time, you define a sequence of steps, choose a harness, and let FlightPath execute the work while keeping outputs, session history, and validation results together.

It ships in two forms:

Surface	What it is
VS Code extension	Dashboard UI, prompt editor, embedded terminal, schema validation, and harness orchestration inside VS Code
`flightpath` CLI	Headless or interactive workflow execution from a terminal or CI job
`flightpath-mcp`	Standalone MCP server used for queue inspection and queue extension

Where to start

New here: Installation and Quick start
Writing workflows: Workflow overview and YAML format
Choosing a harness: Harness overview
Looking up fields: Workflow schema reference

Current documentation scope

This first docs pass focuses on the public surface that is most useful during setup and authoring: installation, workflow syntax, harness selection, key settings, CLI usage, and troubleshooting. Deeper per-harness and per-command coverage can expand without changing the site structure.