ServicesCoding agent tools

Claude Code rollout, starting with one pilot team

We set up Claude Code around your codebase, permissions and review process, and run a scoped pilot with one team before anyone else joins.

Talk to an engineer

.claude/settings.jsonExample

{
  "permissions": {
    "allow": [
      "Bash(pnpm test:*)",
      "Bash(pnpm lint)"
    ],
    "deny": [
      "Read(./.env)",
      "Bash(curl:*)"
    ]
  }
}

Preparing your repositories

We begin where your engineers already are: their repositories and their current tasks. We write instructions with you that explain how the project is built and tested and how changes are documented. That way Claude Code can follow the same steps as your team. We also settle which tools and commands it may run.

A CLAUDE.md listing the commands that really build and test your project
Allowlists for shell commands, file edits and network access, plus rules for when a person has to approve
Hooks that call your existing checks each time the agent has changed a file
Documentation conventions that match what your reviewers ask for today
Which data and which model endpoints are in scope, agreed in advance

Running the pilot

Together we pick a handful of typical tasks, plus a few engineers who will work on them. Their changes go through your normal review and checks. We report results for each team, task and model, and never for a single person.

The pilot results decide which teams follow, and when.

How consulting and training fit together

Consulting gets Claude Code working inside your engineering system. That covers configuration, data flows, permissions and CI, plus a pilot with measured results. Training then builds on that groundwork, with engineers practising on tasks from their own backlog. A team needs both, and the setup comes first.

Claude Code training · Coding agent adoption

Who owns which part of the configuration

Claude Code takes its project instructions from CLAUDE.md, while permissions and hooks live in settings files. On top of that it can load Skills, subagents, plugins and MCP servers. Any of these can belong to one engineer or be shared by everyone. With you, we decide what goes into the repository, what administrators set centrally in managed settings and what each engineer keeps personal.

Inside each repository: CLAUDE.md and project settings, with changes reviewed like any pull request
Rules for the whole organisation, enforced through managed settings
Skills and subagents for tasks that come up again and again, each with a person responsible for it
An allowlist of MCP servers, each with credentials limited to its own purpose

Claude Code in scripts and CI pipelines

Claude Code can also be launched by a script or by CI, with no one following each step as it happens. We look at that only after the pilot, as a separate decision. Running this way calls for narrow task types and an isolated environment. It also needs credentials that cover only the job and a review step the agent cannot change. The decision to push or merge sits in pipeline code, never with the agent.

Our guide to coding agent security

What stays with your team

The CLAUDE.md and settings files for each of your repositories
Runbooks that let your engineers carry on without us
The numbers from the pilot
On request, a monthly session on changes in Claude Code and in the models it uses

What we measure during the pilot

The cost of each accepted change, compared with the time before the pilot
The share of Claude Code pull requests that reviewers send back for another pass
The number of merged changes each team made with Claude Code

Questions

Do all our teams have to start using Claude Code?

No. You can run the pilot on a single repository and then decide how to continue.

Do you also train our engineers?

Yes. Claude Code training takes place in your repositories, once the setup is done.

Can we buy Claude licences through you?

No. Claude plans and API access come from Anthropic, either directly or via your cloud provider. We take no money from vendors and stay independent.

Can your engineers come to our offices?

Yes. We work at your premises in Poland and Germany, or remotely. Our working languages are German, English and Polish.

Related services

Claude Code training

Workshops where your engineers practise Claude Code on their own repositories and backlog tasks.

Harness engineering

The setup that tells an agent how to build, test and change your code, kept in version control.

Evaluations on your repositories

Tools, models or settings tested side by side on changes you have already merged, before you commit to one.

From the blog

Claude Code team instructions

How a team assigns owners to CLAUDE.md files and rules, reviews them, and retires entries that no longer change what Claude Code does.

Coding agent security

Practical security and governance for coding agents: prompt injection, secrets, permissions, sandboxing, unattended runs and ownership.

Talk to an engineer

Bring one piece of work you would like to improve. The call takes 30 minutes and needs no access to your systems or your data.

Book a 30-minute call

Booking runs on Google Calendar.

Prefer to write first? Leave your email and a few lines, and we reply with times.