Define Intent
Author your requirements and decisions in plain Markdown. This creates a human-readable "North Star" that your AI assistant can follow without ambiguity.
Cartopian Protocol
Filesystem-first governance for AI-native development
Cartopian gives AI coding agents a project governance protocol made of plain markdown and directory-as-status conventions. No database, no SaaS dependency, no mandatory tooling; just repo-native context for planning, execution, decisions, and review.
Hero visual placeholder
THE GOVERNANCE LOOP
Plan. Govern. Act.
Cartopian mediates the relationship between human intent and machine execution, using the filesystem as a durable source of truth that prevents project drift.
Protocol Governance
Cartopian's filesystem-first protocol tracks every task and decision. The protocol acts as the source of truth, preventing hallucinations and scope creep.
Agent Execution
Your AI agent acts as a project operator, performing the work and updating the protocol as it goes. You retain control; the AI handles the heavy lifting.
The Cartopian Protocol in Action
From Scaffold to Logic: A Worked Example
Follow the development of a simple Python to-do CLI as it moves through the Cartopian lifecycle. See how the protocol maintains state and clear planning even during core implementation.
Command: Pickup Task
$ cartopian session open
$ cartopian task pickup 02-core-list-ops
> Task started: implement core list operations
Project State: Start of Task
Command: Close Task
$ cartopian task close
> Task 02-core-list-ops complete.
> Added: todo/core.py, tests/test_core.py
> Updated: STATE.md, PLAN.md
Project State: End of Task
FR-014 engagement paths placeholder
Engagement paths content zone
Placeholder for contribution, feedback, and contact routes.