Skip to content

aware-aeco/aware

Repository files navigation

AWARE

AWARE is what comes after software-as-product. Apps are text — files you can read in Notepad. AI is the runtime that executes them. Open source is what the format does automatically: there's nothing else to a "proprietary" app once it's plain English. No vendor, no installer, no walled garden. AECO is the wedge; the substrate is universal.

the statement. The five structural truths it rests on are in the decalog.


60-second install demo

$ npm install -g @aware-aeco/cli                       # any OS — recommended
  # alternatives that work today:
  #   curl ... install.sh | bash             # Linux / Mac, no Node
  #   iex (irm ... install.ps1)              # Windows PowerShell, no Node

$ claude-code                    # or codex, or opencode
  ✓ plugin: aware-aeco · 11 agents · 58 curated + 133 raw skills

> Watch this Tekla model. When a welded assembly appears,
> upload its drawing to my Trimble Connect fab folder.

  → composed [Tekla Watcher] → [Welded Filter] → [TC Uploader]
  → wrote ~/.aware/apps/welded-to-tc.flo

$ aware app run welded-to-tc

That's the whole thing. One sentence in your terminal, one plain-text file, one command to run.

winget install aware-aeco and brew install aware-aeco are queued. They need an MSI installer (Windows) and a Homebrew formula (Mac) plus a code-signing cert, all tracked as follow-up phases. The npm + curl-pipe + PowerShell paths above are what works today.


What's in the repo (v0 snapshot)

aware-aeco/
├── 00-vision/              # decalog · manifesto · positioning
├── 10-core/                # agent-spec · app-spec · runtime contract
├── 20-agents/              # 7 agents · 58 skills · all Apache 2.0
│   ├── _core/              #   meta-primitives
│   │   ├── aware-agent-builder/      # any software → an agent
│   │   ├── aware-skill-builder/      # author / port / modify / eval skills
│   │   └── html-report/              # utility: render self-contained HTML output
│   └── aeco/
│       ├── engineering/
│       │   ├── tekla/                # 31 curated skills · stateful · .NET-mandated
│       │   ├── tekla-2025/           # 2,999 cmds · NuGet ref/ reflection (Tekla.Structures.Model 2025.0.0)
│       │   └── tekla-2026/           # 3,179 cmds · NuGet ref/ reflection (Tekla.Structures.Model 2026.0.3)
│       ├── architecture/
│       │   ├── revit-2025/           # 7,430 cmds · NuGet ref/ reflection (Autodesk.Revit.SDK 2025.0.2.419)
│       │   ├── revit-2026/           # 7,647 cmds · NuGet ref/ reflection (Autodesk.Revit.SDK 2026.0.0.9999)
│       │   ├── rhino-7/              # 5,859 cmds · NuGet ref/ reflection (RhinoCommon 7.38.x)
│       │   ├── rhino-8/              # 6,954 cmds · NuGet ref/ reflection (RhinoCommon 8.31.x)
│       │   ├── grasshopper-7/        # 4,506 cmds · NuGet ref/ reflection (Grasshopper 7.38.x)
│       │   ├── grasshopper-8/        # 5,181 cmds · NuGet ref/ reflection (Grasshopper 8.31.x)
│       │   ├── sketchup-2025/        # 1,684 cmds · YARD docs (SketchUp/ruby-api-docs SU2025.0.3)
│       │   └── sketchup-2026/        # 1,713 cmds · YARD docs (SketchUp/ruby-api-docs SU2026.0)
│       ├── construction/
│       │   └── trimble-connect/      # 7 skills · stateless · REST
│       └── cross-cutting/
│           ├── microsoft-365/        # 4 skills · Graph REST
│           └── google-workspace/     # 4 skills · Drive/Sheets/Calendar/Gmail
├── 30-apps/                # reference apps
│   └── _examples/
│       ├── welded-to-tc.flo          # 3-node linear · canonical demo
│       └── qa-drawings-to-tekla.flo  # 7-node DAG · fan-in + fan-out
├── 40-diagrams/            # Mermaid + Excalidraw views of the substrate
└── 50-research/            # design notes, prior art, competitive analysis

Stats

Count
Curated agents 7 (58 hand-written skills)
Reflected agents 10 (Tekla 2025/2026 + Revit 2025/2026 + Rhino 7/8 + Grasshopper 7/8 + SketchUp 2025/2026 — 673 raw skills · 47,152 commands · auto-generated by aware build agent --from-nuget (NuGet ref/ assemblies) and --from-yard (YARD-rendered docs))
Reference apps 2
Meta-primitives 3 (agent-builder, skill-builder, html-report)
AECO verticals covered engineering · architecture · construction · cross-cutting

Read these in order

  1. 00-vision/decalog.md — the five structural truths (5 min read)
  2. 00-vision/manifesto.md — what AWARE is, why now, how it ships (10 min)
  3. 10-core/agent-spec.md — how to write an agent
  4. 10-core/app-spec.md — how to write an app
  5. 30-apps/_examples/ — two worked apps showing the format end-to-end
  6. 20-agents/_core/aware-skill-builder/ — how to write or port a skill
  7. CONTRIBUTING.md — three ways to contribute, all markdown PRs

Status

Substrate v0: content-complete. Decalog, manifesto, specs, 7 reference agents, 58 production skills, 2 reference apps, two meta-primitives, all hosted under Apache 2.0.

What's still owed for v0.1:

  • The aware CLI binary (the runtime that executes the spec). Spec is detailed enough that the implementation is straightforward — but a real engineering project.
  • Host-plugin generators for claude-code / codex / opencode.
  • Issue tracker, PR review process, contribution badges.

Until the CLI ships, the repo is usable as documentation + reference content: contributors can read the substrate, port their own skills via the documented pipeline, write new agents, and compose apps in .flo format. The first agentic-CLI user who installs this gets a real productivity boost from the skills alone — the AI consults them when composing AECO code.


License

Apache 2.0 — permissive, patent grant, no walled garden. Consistent with the decalog: AWARE's substrate is open by construction.

Commercial apps built on top of AWARE choose their own license. The substrate does not impose one.

FloLess is one such commercial app — a visual canvas for AWARE apps. It is a separate project under its own license.


Watch this repo

The substrate's content is in. The runtime is next. Star the repo to be notified when the CLI ships.

About

Open-source agentic substrate for AECO. Apps are text. AI is the runtime.

Topics

Resources

License

Contributing

Stars

Watchers

Forks

Packages

 
 
 

Contributors