01
Keep the coding agent you prefer
Use Codex, Claude, Copilot, Cursor, Gemini, or another agent. Workspai is the workspace layer around the agent, not a replacement for it.
01open-source workspace intelligence
Give every AI agent the right workspace context.
Workspai maps your projects, dependencies, graph, health, and agent context—then verifies the result through one shared workspace loop.
one model → bounded evidence → verifiable change
$npm install -g workspai
workspace intelligence run --for-agent generic --strict --json
01 Understandmodel + graph current
02 Changediff + impact bounded
03 Evidencedoctor + contracts refreshed
04 Gatereadiness verified
05 Groundagent context prepared
06 Distributeconsumer surfaces synced
07 Explaindecision trail written
receipt.workspai/reports/workspace-intelligence-run-last-run.jsonconsumerspeople · CI · IDEs · AI agents
workspai:/commerce-platform $
02 / choose an outcome
What do you need your AI to do?
Choose the outcome first. Workspai connects it to the right workspace model, graph, skills, health checks, repair flow, or project starter.
workspai outcome resolverselected
Your path
Agent context and skills
Ground Codex, Claude, Copilot, Cursor, Gemini, and other coding agents in the same current workspace evidence.
What you get
- Agent-ready workspace context
- AGENTS.md and tool-specific grounding
- Generated skills and playbook index
03 / command surface
Same workspace intelligence from the terminal
Create or adopt software, inspect the graph, find blockers, prepare agent context, and verify the result from one deterministic CLI. Terminal, CI, VS Code, and AI agents read the same workspace identity and evidence.
$npm install -g workspai
Adopt the current project
Fresh scoped graph search
Bounded Doctor diagnosis
Repair capability truth
Polyglot lifecycle run
Intelligence chain
Or run instantly with npx workspai … — no global install required.
Examples and all mirrored schemas are aligned with Workspai CLI 0.56.0. The extension compatibility floor remains 0.56.0. The complete reference covers 47 live command surfaces and 50 governed artifact mappings.
04 / build + understand
Workspace Intelligence — not a chat window
Workspai turns repositories, projects, dependencies, rules, changes, and evidence into shared understanding for developers, CI, IDEs, and AI agents. Start with a Workspace Intelligence workspace, then bring in the software you already have or create a supported backend, frontend, desktop, or extension project. Workspai derives a workspace-knowledge-graph.v1 graph from the canonical model so CLI, CI, IDEs, MCP clients, and agents can query the same proof-backed system view.
1
Create a workspace
The workspace is the shared architecture around your projects. It starts with a profile-aware README and shared context, without forcing a framework or rewriting your source.
npx workspai my-workspace2A
Bring an existing project
Adopt a local project in place or import a folder or Git repository. Any language or framework can enter the workspace.
npx workspai adopt /path/to/projectnpx workspai import <path|git-url>2B
Create a new project
Choose a supported backend, frontend, desktop, or extension option. Workspai creates it and registers it in the workspace automatically.
npx workspai create projectWorkspai project kits
These backend scaffolds are supplied and governed by Workspai and RapidKit Core, with a 52 modules ecosystem available after creation.
FastAPI Standard Kit
fastapi.standard · Core-backed · stable
FastAPI DDD Kit
fastapi.ddd · Core-backed · stable
NestJS Standard Kit
nestjs.standard · Core-backed · stable
Spring Boot Standard Kit
springboot.standard · Workspai-owned · stable
Go Fiber Standard Kit
gofiber.standard · Workspai-owned · stable
Go Gin Standard Kit
gogin.standard · Workspai-owned · stable
ASP.NET Core Clean Web API
dotnet.webapi.clean · Workspai-owned · preview
Rust Axum API
rust.axum · Workspai-owned · preview
Official ecosystem generators
Workspai invokes each ecosystem's latest stable generator, checks declared runtime requirements, then registers the result and refreshes workspace evidence. The exact commands and requirements come from the CLI's tested create contract.
Next.js· create-next-app· frontendReact Router· create-react-router· frontend· node >=22.22.0React + Vite· create-vite· frontendVue + Vite· create-vite· frontendSvelte + Vite· create-vite· frontendSolid + Vite· create-vite· frontendVite· create-vite· frontendNuxt· create-nuxt· frontendAngular· Angular CLI· frontend· node >=20Astro· create-astro· frontendSvelteKit· sv create· frontendTauri· create-tauri-app· desktopElectron· Electron Forge· desktopVS Code Extension· generator-code· extension· node ^22.22.2 || ^24.15.0 || >=26.0.0Laravel· Composer· backend
Both paths enter the same contract-backed Workspace Intelligence chain: Understand, Change, Evidence, Gate, Ground, Distribute, Explain.
What the intelligence layer produces
Workspace architecture
deterministic layer between code & AIWorkspace
catalog-websaas-apiauth_core.workspai/
Real repos, projects, modules, and kits — not prompts.
Workspace Intelligence
Understand→Change→Evidence→Gate→Ground→Distribute→Explain
System understanding
Governance evidence
Agent grounding
Explainable decisions
Understanding, evidence, grounding, and explanations from one contract-backed chain.
Surfaces
Workspai
VS Code
Workspai CLI
npx workspai
Copilot · Cursor
grounded chat
CI pipeline
--strict gates
05 / capabilities
One controlled loop from issue to verified change
Workspai keeps every workspace action inside the same operating model: detect the real context, diagnose with evidence, plan the smallest safe move, verify before trust, and retain what the team learned. This nested Studio workflow returns to the contract-defined chain through workspai workspace intelligence run --for-agent generic --strict --json.
Workspace platform
Free · No AI requiredWorkspace Explorer
Dedicated sidebar tree showing every workspace, project, and RapidKit Core module. Expand, right-click, or run quick actions without leaving VS Code.
Workspaces · Projects · Modules
Project Scaffolding
One command creates a guided starter project for supported stacks. Pick FastAPI, NestJS, Go / Fiber, Go / Gin, Spring Boot, ASP.NET Core, Frontend Apps, Rust / Axum, Desktop Apps, VS Code Extension, Laravel and Workspai generates the core directory structure, config, and entry point for you to review.
FastAPI · NestJS · Go / Fiber · Go / Gin · Spring Boot · ASP.NET Core · Frontend Apps · Rust / Axum · Desktop Apps · VS Code Extension · Laravel
Module System
Browse the RapidKit Core module catalog for FastAPI/NestJS projects, install with a click, and manage per-project dependencies. Go / Fiber, Go / Gin, Spring Boot, ASP.NET Core, Frontend Apps, Rust / Axum, Desktop Apps, VS Code Extension, Laravel use scaffold-first workflows without the module catalog.
FastAPI/NestJS modules · Advisor suggestions at create
Health Doctor
Versioned workspace and project diagnosis with stable findings, causal groups, proof bindings, explicit unknowns, and governed repair posture.
6 domains · Causes · Proof · Receipts
Verified Engineering Goals
Turn release, dependency-security, and coverage requests into durable definitions of done shared by CLI, CI, IDEs, and agents.
Release · Security · Coverage
Dev Server Control
Start, stop, and restart project dev servers from the VS Code sidebar through each project's effective runtime capabilities.
One-click run · Live status · Logs
Workspace Dashboard
Rich webview panel showing project tree, health status, module inventory, and quick-launch AI flows — the single pane of glass for your governed workspace.
Status · Navigator · Studio/Advisor entry points
Knowledge Graph & Contracts
Query the model-derived Workspace Knowledge Graph, follow portable proofs, and verify the contract registry from the same evidence generation.
Search · Evidence · Path · Verify
Archive Handoff
Export, verify, hydrate, and share workspace archives with integrity checks. Supports local handoff bundles and remote archive import with doctor evidence.
Export · Verify · Hydrate · Share
Studio and Advisor capabilities
Free · Studio/Advisor sessions · CLI agent-syncWorkspai Studio
Nested Detect → Diagnose → Plan → Verify → Learn workflow for workspace incidents. It returns to workspai workspace intelligence run --for-agent generic --strict --json for the complete system gate.
Guided Create
Describe the product or pick a kit. Workspai plans the workspace/project target, then scaffolds only after confirmation.
"Build a SaaS billing API with auth, Stripe, and Postgres"
Project Assistant
Context-aware Q&A grounded in your actual project files, modules, and architecture. Answers that fit your specific codebase.
"What breaks if I change this settings model?"
Workspace Advisor
Advisor sessions for workspace, project, card, and editor scopes. Ask, explain, or navigate while keeping execution user-controlled.
"Explain the auth module in this project"
@workspai in Chat
Native VS Code Chat participant. /ask for architecture questions, /debug for a structured Root Cause → Fix → Prevention response using live diagnostics.
"@workspai /debug Why does this endpoint return 500?"
Editor Issue Actions
Editor lightbulb actions for ordinary code diagnostics: Fix with Workspai opens Studio, Explain with Workspai opens Advisor.
"Why does this endpoint return 422 Unprocessable Entity?"
Fix by Workspai
Artifact cards open Studio sessions backed by remediation plans and approval-bound CLI repair transactions, typed decisions, exact-target verification, and rollback.
"Fix missing environment variable in billing-platform"
Explain with Workspai
Ask Advisor to explain an editor diagnostic, card blocker, or workspace question without applying changes.
"Preview the safest fix for this 500 error"
Change Impact
Use Workspai impact evidence to understand blast radius, risk, and verification targets before making a change.
"If I change this auth middleware, what might break?"
Terminal Evidence Review
Send terminal output, stack traces, or failing test logs into Advisor/Studio as evidence for scoped diagnosis.
"Analyze this pytest failure and suggest the safest next step"
Agent Context Pack
Inspect generated workspace context and grounding surfaces consumed by Copilot, Cursor, Claude, Codex, Gemini, CI, and Studio; the canonical runner owns normal regeneration.
"Always use async SQLAlchemy and avoid sync session in routes"
Agent Sync
Targeted recovery for portable agent grounding drift; normal regeneration remains a stage of the canonical Workspace Intelligence runner.
"Set API naming conventions and architecture decisions"
Agent Skills & Playbooks
Generated operational skills/playbooks for diagnosing, repairing, and verifying workspace issues from evidence.
"Run ship-readiness recipe for this workspace"
Module Advisor
AI scoped to each RapidKit Core module — install guidance, usage patterns, and config tips. Suggests the right modules during project creation.
"How do I wire up db_postgres with auth_core?"
Command Evidence History
Inspect recent command outcomes, reports, and artifact freshness so Studio and the dashboard stay evidence-based.
"Show telemetry summary for the last 7 days"
Coming soon
SoonPro
Module Generator
Add feature slices to any existing project. Routes, services, schemas, migrations, and tests — all in one step with a preview diff.
"Add OTP login with Redis rate limiting and tests"
SoonPro
Studio Debugger (Advanced)
Paste logs, traces, or test failures. Workspai analyzes root-cause patterns and proposes a concrete remediation path in context.
"Why is this endpoint returning 500 in staging?"
SoonPro
Test Generator
Generate unit and integration tests for existing code, covering happy paths, edge cases, and missing surfaces.
"Generate tests for the booking cancellation endpoint"
SoonPro
DevOps Assistant
Generate Docker configs, CI pipelines, env validation, and deploy checklists from your actual project structure.
"Generate a deploy-ready Compose starter for this API"
SoonPro
Architecture Advisor
Architecture guidance grounded in your workspace — caching, queue systems, modularisation, and scale tradeoffs.
"How should I scale this service for 10x traffic?"
SoonTeam
Team Workspace Memory
Shared workspace conventions, code style, and architecture decisions synchronized by Workspai across your whole team.
"Enforce our naming conventions across all new modules"
06 / repair state machine
From detection to learning
in five phases
This is the nested Studio incident workflow, not a replacement for the Workspace Intelligence chain. Each phase has a clear exit condition; after a repair, workspai workspace intelligence run --for-agent generic --strict --json revalidates the complete contract order for CLI, CI, IDE, and agent consumers.
✓
Detect2
Diagnose3
Plan4
Verify5
Learn01Detect
Surface the problem or intent
Workspai reads your workspace — projects, modules, contracts, Doctor evidence, and model changes — so analysis starts from current, inspectable context.
Kitfastapi.standard
Modulesauth_core · db_postgres · stripe_payment
Doctor1 warning — STRIPE_WEBHOOK_SECRET missing
Git diff+routes/payments.py · +tests/
02Diagnose
Root-cause the issue with evidence
Workspai Studio analyzes the detected signals and returns a scoped diagnosis — probable root cause, affected files, confidence, and supporting evidence.
Diagnosing
Root causeSTRIPE_WEBHOOK_SECRET absent in staging env
Affectedroutes/payments.py · stripe_payment module
ConfidenceHigh — doctor evidence + import trace
03Plan
One safe next action
Workspai proposes the smallest correct next step — not a list of options, but one scoped action with explicit impact context and a preview before any change.
Recommended action
Add STRIPE_WEBHOOK_SECRET to .env.staging
·Preview fix before apply
·Scope: staging env only
·Risk: low — config change, no code mutation
04Verify
Prove the fix worked
Every shipped guided action that touches code or config has an explicit verify path. Run the suggested check and Workspai confirms whether the issue is resolved or escalates.
Terminal
$pytest tests/test_payments.py::test_webhook_signature
✓1 passed in 0.42s
✓Verify: passed — incident resolved
05Learn
Retain what worked
Resolved incidents feed the workspace evidence loop — reports, agent context, and generated playbooks keep future sessions grounded in what was verified.
.workspai/reports/agent-customization-pack.json
Contextworkspace-context-agent.json refreshed after verify
Skillsrapidkit-release and repair playbooks indexed
Tracedoctor -> remediation plan -> verify evidence recorded
VS Code Chat Participant
@workspai lives inside
VS Code Chat
No extra window. No copy-paste into a browser. Type @workspai in the native VS Code Chat panel and get answers grounded in your workspace evidence — project structure, installed modules, git diff, reports, and agent context.
/askArchitecture Q&AProject-aware answers grounded in your actual files and modules — not generic LLM output.
e.g.What breaks if I change the settings model?
/debugRoot Cause → RemediationStructured debug flow: Root Cause · Remediation · Prevention. Reads your live VS Code diagnostics as context.
e.g.payments route returning 500 in staging
Chat
Copilot Edits
Context:my-workspacebilling-api
@workspai /debug payments route returning 500 in staging
W
Reading workspace context...
Root causeSTRIPE_WEBHOOK_SECRET missing in staging env
FixAdd key to .env.staging, redeploy
Verifypytest tests/test_payments.py::test_webhook
PreventionAdd env var check to Doctor config
Full context from billing-api · stripe_payment module · .env.example
Ask
@workspai /ask ...
08 / agent neutral
Your agent writes code. Workspai keeps it grounded.
Workspai does not compete for the chat box. It gives coding agents a shared view of the workspace and keeps completion tied to evidence the rest of your tooling can inspect.
02
Give it focused workspace context
Projects, dependencies, graph evidence, contracts, skills, and current blockers come from one governed workspace model instead of one oversized prompt.
03
Verify the result outside the chat
Doctor, impact, tests, builds, contracts, and release evidence decide whether the requested outcome is actually complete.
See the published commands and artifacts in the CLI reference.
Roadmap
Make workspace understanding deeper and more reliable.
We prioritize evidence quality, repair completion, and shared agent context before adding another disconnected surface. Exploring items are directions, not committed release dates.
Available now
One workspace intelligence foundation
Open-source CLI and VS Code surface
Live model, graph, and universal Doctor
Reject stale graph inputs, query scoped proof-backed relationships, diagnose causal findings, and revalidate the complete system with workspai workspace intelligence run --for-agent generic --strict --json.
Agent context, skills, and playbooks
Generate portable grounding for coding agents from the same current workspace evidence.
Verified engineering goals
Plan release, dependency-security, and coverage outcomes with durable definitions of done.
CLI-owned repair transactions
Bind approval to immutable plans, checkpoint changes, execute typed adapters, verify exact targets, and close or roll back through one durable CLI protocol.
Strengthening
More depth, fewer uncertain answers
Current engineering focus
Deeper graph and Doctor coverage
Improve cross-language extraction, service relationships, identity reconciliation, and causal diagnosis.
Broader deterministic repair adapters
Expand runtime and framework depth while preserving decision-required outcomes when safe deterministic execution is unavailable.
Cross-platform generator evidence
Continuously validate official and Workspai-owned project generators against their published contracts.
Studio Debugger (Advanced)
Free editor issue actions are already shipped; the advanced Studio surface is being strengthened for deeper multi-step incident diagnosis.
Exploring
Scale the architecture without fragmenting it
Direction, not a release promise
Focused Workspai packages
Extract stable graph, protocol, security, and shared-contract capabilities when their package boundaries are ready.
Team and portfolio intelligence
Share governed policies, evidence, and risk signals across multiple workspaces and teams.
Richer recovery and sandboxing
Preview and verify higher-risk changes in isolated environments before they reach the real workspace.
Workspai CLI · same workspace truth
ready / start here
Give your next AI task the whole workspace.
One workspace. One truth. Humans and AI aligned. Start from the open-source CLI; add the free extension when you want the same evidence in VS Code.
$npm install -g workspai
no account required · local-first evidence · CLI + VS Code