- Claude Code is Anthropic’s agentic coding tool: it reads your codebase, edits files, runs commands and works with git, from the terminal, the IDE, a desktop app or the browser.
- For an IT team, the biggest gains are on the work nobody has time for: tests, bug fixes, refactoring, dependency updates, code review and documentation.
- It is built for companies: permission modes, sandboxing, organisation-wide managed settings, and no training on your code under commercial plans.
- 123 Consultancy resells Claude plans and trains your developers and IT admins to use Claude Code productively and safely.
AI assistants that suggest the next line of code are now common. Claude Code goes a step further: it is an AI coding agent that understands a whole repository, plans a change, edits the files, runs the tests and prepares the commit. Your developers stay in charge, but the repetitive work moves to the agent. Here is what that means for an IT team, and how to roll it out without taking risks with your code or your data.
What is Claude Code?
Claude Code is an agentic coding tool made by Anthropic, the company behind the Claude models. You describe a task in plain language, in English or French, and Claude Code reads the relevant files, proposes a plan, writes the code across several files, runs commands such as the test suite, and shows you every change before it is applied.
It works where your team already works:
- Terminal: the full command-line tool, on macOS, Linux and Windows.
- IDE: extensions for VS Code (and Cursor) and a plugin for JetBrains IDEs, with inline diffs.
- Desktop and web: a desktop app to review changes visually and run several sessions side by side, and a browser version for long-running tasks in the cloud.
- Your pipeline: integrations with GitHub Actions and GitLab CI/CD for automated code review and issue triage.
What your IT team can do with it
Clear the backlog of tedious tasks
Writing tests for untested modules, fixing lint errors across a project, resolving merge conflicts, updating dependencies, writing release notes: the work that is always postponed. A developer can ask Claude Code to write tests for the authentication module, run them and fix the failures, then review the result.
Build features and fix bugs faster
Paste an error message or describe a symptom: Claude Code traces the problem through the codebase, finds the root cause and proposes a fix. For a new feature, it plans the approach, implements it across files and checks that it works.
Understand legacy code
New team members, or anyone opening an old application, can ask questions about the architecture, the data flow or why a function exists, and get answers grounded in the actual code. This is often the fastest win for teams maintaining older systems.
Automate git and code review
Claude Code stages changes, writes commit messages, creates branches and opens pull requests. In CI, it can review every pull request and flag bugs or security issues before a human reviewer looks at it.
Connect your tools
Through the open Model Context Protocol (MCP), Claude Code can read tickets in Jira, documents in Google Drive, messages in Slack or your own internal tools, so the agent works with the same context as your team.
Claude Code reads a CLAUDE.md file at the root of each project: your coding standards, architecture decisions and review checklist. Writing it well is one of the first things we do with a team, because it makes the agent follow your rules instead of generic ones.
Is it safe for company code?
This is the first question IT managers ask, and rightly so: an agent that can run commands needs clear limits. Claude Code provides them, but they have to be configured.
| Risk | Control in Claude Code |
|---|---|
| The agent runs an unwanted command | Permission modes: in manual mode it starts read-only and asks before editing files or running commands; allow and deny rules per project or for the whole organisation |
| Access to files or networks it should not reach | Sandboxing of shell commands with filesystem and network isolation; a working-directory boundary |
| Each developer configures it differently | Managed settings that the organisation enforces and users cannot override, including which MCP servers are allowed |
| Malicious instructions hidden in files or web pages (prompt injection) | Approval of network commands, trust checks on new folders, and human review of changes |
| No visibility on usage | Usage monitoring through OpenTelemetry metrics |
What happens to your code and data?
Under Anthropic’s commercial terms, which cover the Team and Enterprise plans, the API and cloud providers, Anthropic does not use the code or prompts sent to Claude Code to train its models, unless your organisation explicitly opts in to a program that provides data for model improvement. Standard retention for commercial accounts is 30 days, and zero data retention is available to qualified Enterprise customers. Companies that want the traffic to stay in their own cloud can also run Claude Code through Amazon Bedrock, Google Cloud or Microsoft Foundry.
The practical consequence: use company accounts, not personal subscriptions, so that your code is covered by the commercial terms and your admins keep control.
How to roll out Claude Code in your IT team
- Choose the plan: Claude Team or Enterprise seats for your developers, or API access through your cloud provider.
- Set the rules first: managed settings, permission rules, sandboxing and the list of allowed MCP servers, defined by IT before the first developer starts.
- Start with a pilot: a few volunteers, a few repositories and clear tasks: tests, bug fixes, documentation.
- Train the team: how to give good instructions, write the
CLAUDE.mdfile, review an agent’s work and recognise its limits. - Measure and extend: time to fix bugs, test coverage, pull requests merged, developer feedback after 30 and 90 days.
How 123 Consultancy can help
123 Consultancy is a reseller of Claude plans and a hands-on AI training company. We can provide your licences and set up Claude Code with your IT team, then train your people to use it well:
- Developer bootcamp: one or two days of practice on your own repositories, from first tasks to multi-file features, tests and pull requests.
- Admin and security workshop: managed settings, permissions, sandboxing, MCP governance and monitoring, so IT stays in control.
- Follow-up: a project
CLAUDE.md, shared team workflows and support during the first weeks.
Sessions run in French or English, on site in Tunisia or France, or remotely. See our AI training programs or tell us about your team: we will propose the right plan and program.
FAQ
What is Claude Code?
Claude Code is an agentic coding tool from Anthropic. It reads your codebase, edits files, runs commands and works with git and your development tools. It is available in the terminal, in VS Code and JetBrains IDEs, as a desktop app and in the browser.
Does Anthropic train its models on our code?
Not under commercial terms. For the Team and Enterprise plans, the API and cloud providers, Anthropic does not train its models on code or prompts sent to Claude Code, unless your organisation explicitly opts in to a data-sharing program.
Can IT control what Claude Code is allowed to do?
Yes. Administrators can enforce managed settings that users cannot override: permission rules, allowed commands and MCP servers, sandboxing of shell commands and usage monitoring.
Does Claude Code replace developers?
No. It takes over repetitive and time-consuming tasks and speeds up the rest, but developers decide what to build, review every change and stay responsible for the code that ships.
Can 123 Consultancy provide the licences and the training?
Yes. 123 Consultancy resells Claude plans and trains developers and IT administrators to use Claude Code productively and safely, in French or English, on site or remotely.
