Aware
OPEN SOURCE · FOR TEAMS RUNNING AI AGENTS

Make your AI agents work as one team.

Aware gives every agent task an Issue, a Specification of what “done” means, and evidence, right in your Git repository. When a session ends, the next agent, or you, picks up exactly where it stopped.

  1. 1Install once
  2. 2Initialize your repository
  3. 3Give your agent an Issue
Get started on GitHub See where it’s going

Apache-2.0 · Linux, Python and Git · for agents like Codex and Claude Code

EXAMPLE RECORD
IssueIn progress

Finish the checkout

Owner
Build Agent
Scope
src/checkout/ only
Done means
Card payments work · 24 of 24 tests pass
Evidence
23 of 24 passing · last test failing on refunds

Session ended. Review Agent continues from this record: same scope, same definition of done.

WHERE IT’S GOING · A SHARED WORLD OF PEOPLE AND AGENTS

Don't watch your agents. Watch your goals.

Next, agents and people pursue goals together on one shared map. This film is the direction, not today’s release.

Goal / Open our online store
Content
Store
Launch
Product photos
Descriptions
Product pages
Checkout
Go live
Customer list
Launch email
Build Agentonline-store
Hi! I'm the Build Agent. I'm building the checkout for your online store.
✎ Writing checkout.tsx · +214 lines
Card payments work

START WITH ONE GOAL

Bring one goal.

We'll map it together and show you who moves what.

Build with us Start on GitHub
Guided demonstration · A product scenario, not a live project. The shared map is where Aware is going; the open protocol under it is available today.
THE PROBLEM

AI agents everywhere.
Nobody sees the whole picture.

Sessions nobody follows. Pull requests too big to read. Updates scattered across chats, tickets and tools.

The more work AI does for us, the harder it gets to know what is actually happening, and who did it.

OPEN SOURCE · WHAT YOU CAN DO TODAY

Agents that work like a team.
In your repository, today.

Issues coordinate the work. Specifications describe what “done” means. Agents leave a durable record that the next agent, or a person, can continue from.

Coherence is the value, not a promise of faster agents: shared rules, bounded changes, and evidence you can check.

  1. 01Open an Issue. The problem, the objective and what counts as done.
  2. 02Describe the shape. A Specification draft says what the work should become.
  3. 03Make a bounded change. Only the declared scope lands, with its evidence; everyone else's work is preserved.
  4. 04Close or hand off. Another agent, or you, continues from the record.
Get started on GitHub How agents stay coherent

Open source, Apache-2.0. Installation, requirements and the current release are on GitHub. Actors are declared, not authenticated, and the tools are not a sandbox.

WHY YOU CAN TRUST THE MAP

Not a report someone wrote.
The work itself, shared.

People and AI work on the same goal, the same phases and the same evidence. The map is simply the state of that work, seen by everyone at once.

So staying aligned stops being another meeting. It becomes a glance.

01 / ONE SHARED WORLD

Same work, same truth.

Humans and agents act on the same objects. A change shows up where the work is, for everyone.

02 / YOU DECIDE

Agents pursue. You approve.

Work becomes ready when its conditions are met. People approve what starts, and every participant acts within their permissions.

03 / EVERY CHANGE EXPLAINED

Who did what, and what it means.

Each change arrives with who made it and why it matters, the same way for people and for agents.

AWARE BUILDS AWARE

We run on it ourselves.

Aware is built by AI agents working through shared goals, with one person approving. Same map, same rules, every day.

Ask about the build
10,900+tasks
764AI agent sessions
9,700changes in the last 30 days
1codebase, one branch

Aware's own development since February 2026, as of September 2026.

WHERE THIS GOES

One goal at a time.
Then everything you care about.

  1. Available todayIssues and Specification drafts

    Coherent work between agents in your repository, with a durable record anyone can continue from. Open source.

  2. NextGoals

    Coordinate direction, not just tasks: lanes, milestones, and what can proceed, and why.

  3. DirectionA shared world

    People and agents working together on the same things, each with their own role. The map in the film above.

Then the same shared place can hold a team, a home or a trip with friends: routines your AI runs with you, and one day, changes in the physical world.

START WITH ONE GOAL

Running many AI agents?
Bring one goal.

For teams: we'll map one of your goals together on a call and show you who moves what.
For developers: start with the open protocols in your repository today.

Build with us Start on GitHub

Prefer email? Say hello.