schema: https://brainfile.md/v1

The protocol AI agents can parse. The format humans can read.

// No databases. No lock-in. Just text.

brainfile.md

# YAML frontmatter
---
schema: https://brainfile.md/v1
columns:
  - id: in-progress
    tasks:
      - id: auth-1
        title: Implement OAuth 2.0
        priority: high
        tags: [backend, security]
        subtasks:
          - id: auth-1-1
            title: Set up credentials
            completed: true
---

# Markdown content
Additional context for AI agents.

What AI sees

✓

Schema validated

v1 spec enforced

✓

Tasks parsed

IDs preserved, structure intact

✓

Rules extracted

Agent instructions ready

✓

Context loaded

Markdown content available

Format

YAML + Markdown

Storage

Git-tracked text

Agents

Read & modify

How agents use this

Drop brainfile.md in your repo. AI agents can read it, understand the structure, and follow your rules.

Claude / Cursor

"Read brainfile.md and move task-5 to done. Update the completed timestamp."

GitHub Copilot

"Add a high priority task for fixing the login bug to the todo column."

Any LLM

"Parse the YAML, validate against schema, extract all tasks with priority=high."

Your scripts

cat brainfile.md | grep 'priority: high' | wc -l

Why this exists

Working with AI agents on tasks is messy. Context is spread across tools. Changes aren't tracked. Brainfile is a file you commit. Tasks live in your repo. Agents follow explicit rules. Full git history.

Official tools (optional)

@brainfile/core

npm install

Parse, validate, and serialize

@brainfile/cli

npm i -g

Command-line interface

VSCode Extension

marketplace

Visual kanban board