Get Started with the Dynapps AI Platform
Choose your Dynapps client and follow the platform setup, sign-in, and verification guides.
On this page
Connect your AI client to your organization's gateway to use the models, skills, and tools your account can access. You need an account and Dynapps Bridge on your computer; you do not need to build or run a server.
Choose your client
| What you want to use | Setup guide | Platforms |
|---|---|---|
| Claude Code in a terminal, including gateway Gemini models | Connect Claude Code | macOS, Linux, native Windows; optional WSL |
| Claude Cowork in Claude Desktop | Connect Cowork | macOS and Windows |
| OpenCode | Connect OpenCode | See the client guide |
Claude Code and Cowork are different clients. Connecting one does not configure the other. On Windows, follow the native PowerShell setup and enrol each client separately. The Windows guide identifies the bridge build needed for Claude Code automation; WSL uses a separate Linux bridge.
Before you start
Ask your administrator for your gateway's HTTPS address, access to an account, and the trusted bridge manifest public key if your device has not already been configured. Use the same gateway for downloading, signing in, and connecting. Do not rely on the app's default address.
Start with Install on macOS, Install on Windows, or the Linux and WSL instructions. Then follow your client guide.
Verify your first session
Ask a short question in the connected client. Check that the request appears in your gateway's History, or ask your administrator to confirm it. A successful sign-in alone does not prove that model requests are routed through the gateway.
Next, try a read-only skill or tool that your administrator has enabled. Model access and tool access are separate: a working model response does not prove that plugins are installed or authorized. Browse the skills catalogue for examples.
Skills
Changing a skill and getting it in front of agents is one walkthrough: Update a Skill and Ship It — edit, gate, publish, import, verify. Maintainers can use the skill lifecycle guide to trace a skill from its source through distribution, loading and measurement.
Sync: code and instance
- Code ↔ Instance: sources, planes and sync — the model behind it: where declarations come from, what they project into, which hash proves what
- Access control: who reaches what, and why — entitlement, declared here and enforced from the database
Engineering
- Agent farm from Claude Code: runs, evals and usage — the agent-farm MCP server for developers and project managers
Analysis: measuring and comparing skills
- Analysis overview: the tabs and the end-to-end path
- Measure which skills are used
- Evaluate a plugin before you ship it
- Create a conversation and see it land
- Versions: hashes, history and compare
- Artifacts and the tool-call ledger
Connections
Administration
- Dashboard — the admin overview
- Create and manage users — including the one role table
- Authentication
- Set up SSO (Microsoft Entra ID)
- Secrets management — secrets sources, 1Password, rotation
- User and access management
- Usage and analytics
- Cost management
- Model routing and data residency
- Gateway API
- Conversation history and search
Governance and compliance
- Governance modes: warn and enforce
- Safety and guardrails
- Tool governance
- Audit and observability
- Export telemetry (OTLP)
- Data retention
- Roadmap and known limitations
Deploy and operate
- Stand up one node and PostgreSQL
- Deploy on Oracle Cloud
- Validate every stage
- Expose your instance remotely
- Target deployment architecture
For platform maintainers: Develop Against a Local Gateway.