How Software Arrives in Your Instance, and What Checks It at the Door
For whoever has to approve it: how modules arrive and install, what an agent is allowed to do, how your AI tools connect, the engine behind it, and the questions a security review asks.
Capabilities Install Like Software Packages
Every capability is a versioned module from the catalog. Installing one brings in everything it depends on and applies it as one recorded change, while the app keeps running. If any step fails, the whole change rolls back.
- A review shows what will install, what it needs from your copy, and which database changes run
- Each module shows its publisher, license, versions and readme before you install it
- A module from outside the catalog installs too, and is marked as unverified
Writing the Software Is Easy. Trusting It With the Business Is Not.
Once software touches customers, money and records, the questions change: who is it acting as, what can it reach, what happens when it fails, and who approves what it sends. KickSide settles them in its foundation, so every module and agent inherits the same rules.
| Question | Scripts and AI apps today | In KickSide |
|---|---|---|
| Who it acts as | Scripts and AI apps todayA script runs with whatever keys someone pasted into it. | In KickSideEvery agent and automation acts as a specific person and can do only what that person could. |
| What it can reach | Scripts and AI apps todayAI-written code runs with open internet access. | In KickSideIt runs isolated inside the platform and reaches only the tools it was given. |
| When access is checked | Scripts and AI apps todayOnce, at login, if at all. | In KickSideOn every action, against who is acting and what they touch. |
| When it fails | Scripts and AI apps todayIt stops overnight and nobody notices until Monday. | In KickSideEach piece restarts on its own, and long jobs resume where they stopped. |
| Who changed what | Scripts and AI apps todaySomeone edited a server, and nobody knows what or when. | In KickSideEvery install and change is recorded and can be rolled back. |
| What goes out | Scripts and AI apps todayThe email goes out the moment the model decides. | In KickSideAnything consequential waits for a person in the inbox. |
Your AI Tools Connect as You, Never More
Each person can issue their own connection for Claude Code, Cursor, Claude Desktop or any other tool that supports MCP, the open standard AI tools use to work with other software. The tool then works inside KickSide as that person, on the real records, with no more reach than they have.
- Each connection carries one person's identity and can be limited to named parts of the copy
- The tools it can use are chosen when the connection is issued, and the app's own access rules still apply
- Connections expire, and each one can be revoked
- KickSide agents can also use tools from other MCP servers
claude mcp add --transport http kickside <your address> --header "Authorization: Bearer <token>"Issue a connection, paste its address and token into this command, and Claude Code works inside your copy.
Underneath It All: the Wippy Runtime
KickSide runs on Wippy, the runtime our team built for software that keeps changing while it runs. It is a single program, open source under MPL-2.0 and public on GitHub, so you can read the code doing any of this.
| You need | The usual answer | In the runtime |
|---|---|---|
| Many small jobs, safely | The usual answerContainers and an orchestrator | In the runtimeIsolated, supervised workers in one program |
| Work that runs for days | The usual answerA separate workflow cluster | In the runtimeWorkflows that resume after a restart |
| Running other people's code | The usual answerSandboxes with the network open | In the runtimeIsolated sandboxes limited to the tools they were given, with R and Python inside |
| Growing past one machine | The usual answerA re-architecture | In the runtimeThe same program across many machines, SQLite to start and Postgres to grow |
The Questions Your Security Review Will Ask
The short answers are below. The full security and privacy overview covers the threat model, controls and status in detail.
It stays on your servers or in your cloud account. Nothing passes through ours unless you choose KickSide Cloud, and then it stays in your own private environment.
Only what you allow. A new agent starts with no access to files, the internet or other parts of the system, and every attempt is checked against your rules when it is made.
Every action is logged: what ran, for whom, against which record. Your compliance team can read it in your own environment without asking us for an export.
No. One company per copy, with no shared database.
You do. You connect your own model provider and pay them directly. We don't see your keys or mark up your usage.
No. Your data, private modules and synced history are not used to train models. The contract wording that commits us to that is being finalized, and we'll send you the current draft if you ask.
Any time. Your data and the record live in your own environment. If we host your copy and you leave, it stays readable for 30 days so you can export it.
A data processing addendum for deployments where we host your data is being finalized. We'll send the current draft on request.
Not yet. We are preparing for SOC 2, with the audit expected in 2027 at the earliest, and we are not HIPAA certified.
Not yet. Modules from our catalog aren't signed yet, so a module from any source installs and is marked as unverified. Treat it the way you'd treat any code from the internet.
Walk Through It With Our Team
Bring the questions from this page: how modules install, what agents can do, how your AI tools connect, and what your security review needs.
This site is protected by reCAPTCHA.
Rather try it first? Install it in two commands.