Workspai
August 16, 2026 · View on GitHub
Give your AI agent the system—not just the repository
Your coding agent should not have to rediscover your architecture in every session. Workspai builds a current, evidence-backed view of your projects, APIs, dependencies, infrastructure, and change boundaries, then gives agents the bounded context they need.
One workspace. One truth. Humans and AI aligned.
npx workspai adopt .
npx workspai workspace intelligence run --for-agent generic
A single run gives developers and tools the same operational picture:
System projects · runtimes · APIs · dependencies · infrastructure
Evidence relationships linked back to canonical proof
Change affected projects and verification targets
Agents bounded context instead of an unfiltered repository dump
Understand the system before an agent changes it.

Get started · See what you get · How it works · Documentation
Workspace Intelligence for software systems
Software is more than a folder of source files. It includes projects, services, APIs, dependencies, infrastructure, documentation, tests, policies, and release evidence.
Workspai brings those scattered parts together, so people and AI tools can understand and work with the same software system:
- See the whole workspace: understand what exists and how it fits together.
- Ask with proof: trace answers back to the files and facts that support them.
- Change with confidence: see impact, run the right checks, and give AI tools focused context.

Start with your software
You do not need to move an existing project. Open its directory and adopt it:
cd /absolute/path/to/project
npx workspai adopt .
Workspai creates or reuses a minimal workspace in the default system location and links the project to it. You can stay in the project directory:
npx workspai workspace intelligence run --for-agent generic --strict --json
This run builds the current system view, checks its evidence, and prepares
shared context for people and tools. Results are saved under .workspai/.
When something is missing or blocked, Workspai reports it instead of claiming
the workspace is healthy. This canonical form gives CI, agents, and other
machine consumers strict gate semantics through the versioned JSON contract;
omit --strict --json for the shorter human-readable first run shown above.
Starting from scratch? Use the guided flow:
npx workspai create
It can create a workspace, scaffold a project, or add existing software.
Give your agent a goal—not an open-ended prompt
Describe the outcome in plain language from the adopted project:
npx workspai goal "Raise test coverage to 85%" --for-agent generic
Workspai turns it into a bounded, evidence-backed handoff:
Intent → project scope → proof-backed context → governed plan → CLI verification
The agent gets a focused objective, not permission to scan or change everything. Workspai keeps approval, verification, and rollback under CLI control. The command prepares governed work; it does not edit source or claim that the outcome is complete.

What Workspai gives you
| Question | Answer from Workspai |
|---|---|
| What is in this system? | A canonical model of projects, runtimes, frameworks, rules, and current evidence |
| How is it connected? | A searchable graph whose relationships link back to proof |
| What changed? | Saved snapshots, differences, and affected projects |
| Is it healthy or ready? | Doctor, analysis, policy, readiness, and verification results |
| What should an AI tool read? | Bounded context, instructions, skills, and MCP-accessible evidence |
| Why is something blocked? | A diagnosis, supporting evidence, and the next verification target |
The evidence index at .workspai/reports/INDEX.json is the simplest entry point
for developers, IDEs, CI, and agents.
How it works
Code · APIs · packages · infrastructure · docs · CI · policies
│
▼
Canonical Workspace Model
│
▼
Evidence-backed Knowledge Graph
│
▼
impact · doctor · verify · context · explain
│
▼
Developers · CI · IDEs · MCP · AI agents
The Workspace Model is the canonical source of truth. The Knowledge Graph is a derived, revision-bound representation of that model. Providers can enrich the graph with files, symbols, APIs, tests, infrastructure, ownership, and proofs, but they do not rewrite the model during the same run.
A missing relationship means not proven by current evidence, not “these projects are independent.”
The complete decision loop is versioned as a contract:
Model → Diff → Impact → Doctor + Contract Verify + Analyze → Readiness
→ Verify → Context → Agent Sync → Explain
The model, graph, and verification chain run locally and do not require an AI API key. AI providers are optional consumers of the same governed context.
One foundation, many consumers
- Developers get clear summaries, proof paths, and next actions.
- CI gets structured JSON and versioned evidence.
- AI agents get focused context instead of an unbounded repository dump.
- IDEs and dashboards read the same model, graph, and verification results.
- MCP clients can query current workspace evidence.
- Graph tools can use JSON, JSON-LD, Mermaid, DOT, GraphML, or GEXF exports.
Go deeper
| Goal | Guide |
|---|---|
| Learn the main concepts | Plain-language glossary |
| Create, adopt, or import software | Creating workspaces and projects |
| Query the graph and inspect proof | Workspace Knowledge Graph |
| Understand the full decision loop | Workspace Intelligence runner |
| Repair a governed blocker safely | Workspace Repair Engine |
| Compile intent into a scope-bound agent handoff | Goal Packs |
| Integrate CI | CI workflows |
| Find a command or flag | Command reference |
| Inspect schemas and artifact ownership | Artifact Catalog |
Packages
The current CLI already includes the integrated Workspace Intelligence capabilities described above.
Shared and graph foundations are being hardened as future independent packages. They are extraction boundaries for existing capabilities, not a list of missing features.
Develop
npm ci
npm run build
npm test
Read the Development Guide, Contributing Guide, and README content contract.
Community
Workspai is an open-source project by Chistiq, the intelligence infrastructure company behind RapidKit and Workspai.
License
MIT. See LICENSE.