namel3ss

AI-native programming language for deterministic, inspectable applications. Deterministic execution with explicit AI boundaries and governed runtime.

_{One .ai file defines records, deterministic flows, explicit AI boundaries, and UI — with inspectable state changes.}

---

Start here

• Quickstart
• First 5 minutes
• What you can build today
• Stability
• Limitations

Try it in 60 seconds: docs/quickstart.md.

Run modes

• n3 run app.ai runs in production mode and renders only user-facing UI.
• n3 run studio app.ai (or n3 run --studio app.ai) runs the same app with Studio instrumentation in the manifest.
• n3 studio app.ai remains available as the dedicated Studio inspector command.
• UI blocks can be marked with debug_only metadata so they render only in Studio mode.
• Custom UI components are opt-in via use plugin "name" and require custom_ui in capabilities.

What you can run today

• Templates: docs/templates.md
• Template contract: templates/
• Demos: src/namel3ss/demos
• Studio (inspection UI): docs/studio.md

RAG support (deterministic)

namel3ss supports production-grade retrieval-augmented generation as a first-class pattern:

• Deterministic ingestion and retrieval
• Answering with mandatory citations
• PDF page preview and exact source highlighting
• Explain mode for auditable selection
• Optional embeddings (runtime config only, no grammar changes)

See the canonical overview: docs/rag/overview.md

Core stable

Core stable is the graduation gate for the language surface, compiler determinism, and safety guards. See the authoritative definition in docs/graduation/core_stable.md.

Get started

Develop from source

Use this for development and testing.

python3 -m venv .venv && . .venv/bin/activate
python3 -m pip install --upgrade pip && python3 -m pip install -e ".[dev]"
python3 -m namel3ss --help

Run with Docker (isolated runtime)

Use this for a clean, repeatable runtime.

docker build -t namel3ss:local .
docker run --rm namel3ss:local n3 --help

Learn / explore (docs & templates)

Use the local Docker image to open Studio for a bundled demo.

docker run -d --name namel3ss_studio -p 7340:7340 \
  -v "$PWD:/workspace" -w /workspace \
  namel3ss:local n3 app/app.ai studio --host 0.0.0.0 --port 7340
docker logs namel3ss_studio
docker rm -f namel3ss_studio

See docs/install-and-run.md for complete install and Studio instructions.

Installation (summary)

Supported paths:

• Install from source (development)
• Docker (local) for an isolated runtime

Full guide: docs/install-and-run.md

Docker quick check:

docker build -t namel3ss:local .
docker run --rm namel3ss:local n3 --help

Install + Verify

Install (source, editable):

python3 -m pip install --upgrade pip
python3 -m pip install -e ".[dev]"

Verify CLI and package:

n3 --version
n3 doc
python3 tools/package_verify.py

Verify embedding (C example, requires C toolchain + cargo):

python3 -m pytest -q tests/embed/test_embed_c_example.py

Browser Protocol

Browser Protocol is defined in docs/runtime/browser-protocol.md.

Docker & CI guarantees

• Docker builds install from local source, not PyPI.
• Docker builds are validated in CI.
• CLI smoke tests are enforced automatically.
• Wheel installs are smoke-tested in release automation.

References:

• Dockerfile
• CI workflow
• Release workflow
• Docker build guard
• Wheel install check

Release & governance

Release invariants (enforced):

• VERSION is metadata.
• Docker builds do not depend on PyPI.
• Publish is gated by CI and guards.
• Canonical sequence: VERSION bump → tests → tag → PyPI publish → docker image → release notes.

References:

• Release readiness
• Release workflow

How it works (high level)

Design guarantees:

• Deterministic execution
• Explicit AI boundary
• Inspectable state and traces
• Governed memory

Runtime guarantees:

• Stable CLI surface
• Deterministic manifests and outputs
• Read-only diagnostics and explain output
• Safe cleanup of runtime artifacts

See docs/trust-and-governance.md.

UX as a Contract

UX behavior is deterministic and explainable through stable manifests and n3 see output.

Guarantees:

• Uploads (progress, preview metadata, async errors)
• Conditional UI (state-gated visibility)
• Reusable UI patterns (compile-time expansion)
• Accessibility by default (roles, labels, keyboard, contrast)

Detailed UX contracts: docs/ui/overview.md

Language Contracts

Language Contracts are defined here:

• docs/language/application-runtime.md
• docs/language/application-data-model.md
• docs/language/backend-capabilities.md
• docs/language/no-dependencies.md
• docs/language/capability-packs.md

Reserved identifiers

If you must use reserved identifiers, escape them with backticks and use n3 reserved to list them. The reserved words list is in docs/language/reserved-words.md.

Documentation index

Getting started:

• Install and run
• Quickstart

Language & grammar:

• Grammar contract
• Application runtime
• Application data model
• Backend capabilities

Runtime & backend:

• Data and migrations
• Observability
• Model training
• Streaming responses
• Performance hardening
• Browser protocol

UI & Studio:

• Studio
• UI DSL

Governance & releases:

• Release readiness
• Stability

Discussions & Design Conversations

GitHub Discussions is the canonical place for architectural and language design conversations. Use Discussions for design questions, trade-offs, and long-form proposals; use Issues for bugs and actionable feature requests.

Reference discussion: namel3ss-Ai#2

Contributing

See CONTRIBUTING.md.

Summary:

• Clone the repo.
• Create and activate a virtual environment.
• Install editable + run tests (python3 -m pytest -q).

Troubleshooting

• CLI not found: follow docs/install-and-run.md.
• Icon registry missing: use Docker or install from source per docs/install-and-run.md.
• Virtualenv confusion: recreate the venv per docs/install-and-run.md.