Workspai Example Workspaces

Real, cloneable examples of Workspace Intelligence for software systems.

Start from a raw workspace profile, create a supported project, or bring an existing project into the architecture through import or in-place adoption. Languages, frameworks, repositories, and AI tools stay yours; Workspai adds the shared model, evidence, governance, and agent grounding around them.

Some runnable examples were originally generated with RapidKit Core and retain their legacy metadata as compatibility input. Their current workspace interface and all new commands use Workspai.

Validated interface baseline: workspai@0.55.1. The legacy generated projects remain reproducibly pinned to rapidkit-core==0.6.0 until they are regenerated and their module/runtime locks are validated together. Each runnable project carries its own reproducible lockfiles and module registry; Workspace Intelligence remains the shared architecture across them.

Why Workspace Intelligence

Workspai is not another AI coding assistant, agent framework, or replacement for the tools your team already uses.

One workspace. One truth. Humans and AI aligned.

It adds a shared, evidence-backed intelligence layer around your software system. Projects, dependencies, policies, changes, health evidence, and release decisions become one workspace model that developers, CI, IDEs, and AI agents can use together.

You can keep your existing languages, frameworks, repositories, models, and AI tools. Workspai generates portable context and grounding surfaces such as AGENTS.md, skills, structured reports, and tool-specific instructions from the same verified workspace evidence. It complements Claude, Codex, Copilot, Cursor, and other consumers instead of competing with them.

Every workspace in this repository already publishes its initial portable consumer views. Open AGENTS.md for Codex and other agents, .github/copilot-instructions.md for Copilot, CLAUDE.md for Claude, or the rules under .cursor/ and .claude/. These files are projections of one Workspace Intelligence model—not separate hand-maintained instructions. Live reports are rebuilt locally after clone so stale or machine-specific evidence is never presented as current.

Start here

git clone https://github.com/chistiq/rapidkit-examples.git
cd rapidkit-examples
npm run hydrate:core -- \
  --workspace quickstart-workspace \
  --project product-api \
  --project ecommerce-api
cd quickstart-workspace

npx workspai workspace sync
npx workspai workspace contract verify --strict --json
npx workspai workspace intelligence run --for-agent generic --strict --json

workspace sync is idempotent. On a newly cloned machine it registers the workspace locally, discovers its projects, and refreshes the portable workspace contract.

hydrate:core installs the workspace-pinned Python engine and restores ignored module runtime payloads from each project's committed registry.json. When the pinned Core version supports it, hydration uses the locked restore contract:

rapidkit modules restore --locked --ci

If the locked resolver cannot reproduce a project, the wrapper falls back to registry replay and accepts the result only when every restored module still matches the committed version exactly. The user-facing command stays stable and the fallback cannot silently upgrade or downgrade an example.

Choose a learning path

You want to...	Continue with...
Run the smallest complete example	Quickstart Workspace
Start from an empty language/runtime boundary	Raw profile fixtures
Add an existing repository without changing its framework	Import or adopt
Understand changes, impact, and verification	Workspace Intelligence chain
Repair a Doctor or governance blocker safely	Governed repair transaction
Ground Claude, Codex, Copilot, Cursor, or another agent	Agent and IDE files
Add strict release checks	CI and release

Published workspaces

Workspace	Projects	Focus	Level
Quickstart	2	FastAPI, auth, PostgreSQL, Redis, observability	Beginner
AI Agent	2	FastAPI and NestJS agent implementations	Intermediate
SaaS Starter	4	Multi-service SaaS APIs, admin, webhooks	Advanced

Machine-readable catalog: examples.json.

Publication rules: PUBLICATION_CONTRACT.md.

Complete clone, registry, import, adopt, and Workspace Intelligence workflow: WORKSPACE_ONBOARDING.md.

Raw profile fixtures

Every supported profile is published as an empty, CLI-generated workspace at the repository root so its foundation is visible and directly cloneable:

Profile	Raw workspace
minimal	minimal-workspace
java-only	java-only-workspace
python-only	python-only-workspace
node-only	node-only-workspace
go-only	go-only-workspace
dotnet-only	dotnet-only-workspace
polyglot	polyglot-workspace
enterprise	enterprise-workspace

They demonstrate profile and foundation contracts only. Application projects are intentionally kept in the real examples above, while the complete kit matrix is verified by CLI generator tests.

Profile boundaries and publication rules: PROFILE_WORKSPACES.md.

Workspace Intelligence

flowchart LR
    INPUTS["Projects · Dependencies · Policies · Changes"] --> WI["Workspace Intelligence<br/>Model · Impact · Verify · Context"]
    WI --> EVIDENCE["Evidence<br/>Reports · Contracts · Gates"]
    WI --> AGENTS["Agent grounding<br/>AGENTS.md · Skills · Context"]
    EVIDENCE --> CONSUMERS["Developers · CI · IDEs · AI agents"]
    AGENTS --> CONSUMERS

Loading

Each published workspace includes:

• .workspai-workspace as the canonical workspace marker
• .workspai/workspace.json for workspace identity and profile
• .workspai/workspace.contract.json as the portable project contract
• .workspai/policies.yml and .workspai/toolchain.lock for governance
• legacy .rapidkit* metadata required by the original Core-backed projects
• project-level .rapidkit/modules.lock.yaml files for Core module restore

The canonical command for refreshing model, graph, health, contracts, readiness, verification, and agent grounding in contract order is:

npx workspai workspace intelligence run --for-agent generic --strict --json

Its durable receipt is .workspai/reports/workspace-intelligence-run-last-run.json. Individual commands remain useful for focused inspection, but they do not redefine the canonical chain.

The legacy files are compatibility inputs, not the current public interface. Do not delete them until the example projects are regenerated by a compatible Workspai-owned kit. Machine-local reports, registry summaries, and absolute adoption records are intentionally ignored and regenerated after cloning.

Run an example

Quickstart

cd quickstart-workspace/product-api
cp .env.example .env
npx workspai init
npx workspai dev

AI Agent

cd my-ai-workspace/ai-agent
cp .env.example .env
npx workspai init
npx workspai dev

NestJS variant:

cd my-ai-workspace/ai-agent-nest
cp .env.example .env
npx workspai init
npx workspai dev -p 8013

SaaS Starter

cd saas-starter-workspace
npx workspai doctor workspace
npx workspai workspace run init

See each workspace README for service ports, infrastructure, endpoints, and module-specific setup.

Pro showcases

pro-showcase contains public product descriptions only. Paid source code, customer archives, entitlement logic, and release evidence remain private. Showcase directories are not presented as runnable workspaces.

Publication policy

• Free examples must remain cloneable and useful without private dependencies.
• Published commands must use the current workspai npm CLI.
• Portable contracts may be committed; machine-local evidence may not.
• Pro availability claims require a real release candidate or published product.
• examples.json, workspace contracts, and directory contents must agree.

Run the repository integrity check before publishing:

corepack npm run check

Ecosystem

Surface	Link
RapidKit Labs	getrapidkit.com
Workspai product	workspai.com
Knowledge portal	workspai.dev
Workspai CLI	chistiq/workspai
VS Code extension	chistiq/rapidkit-vscode
RapidKit Core	chistiq/rapidkit-core