Workshops: Agentic Coding Tools in Depth

Cline, Roo Code & Continue: Open-Source Coding Agents Workshop

Cline, Roo Code and Continue turn VS Code into an agentic coding environment without locking you into one vendor: you pick the model, connect your own API key or company gateway and see what every task costs. The workshop teaches developers the daily routine and gives your platform team a setup it can govern company-wide.

Who is this workshop for?

  • Developers who work mainly in VS Code (or in JetBrains IDEs, in the case of Continue)
  • Platform and developer experience teams that provide AI tooling to engineers
  • Teams that must route all model traffic through an approved gateway or their own cloud tenancy
  • Engineering leads who want to see cost per task rather than only flat seat licences
  • Organisations that prefer open-source tools they can inspect and adapt

What you will take away

  • Install and configure Cline, Roo Code and Continue and know which job each of them does best.
  • Connect suitable models, from Claude via the Anthropic API, Amazon Bedrock or Google Vertex AI to OpenAI, Mistral or a local endpoint.
  • Read token usage and cost per task, and keep spend predictable with modes, context limits and deliberate model choice.
  • Set up a central LLM gateway with per-team keys, budgets and logging that every agent routes through.
  • Define approval rules, rules files and MCP connections so the agents act in a controlled way in your repositories.

Agenda

  1. Module 1

    The open-source agent landscape

    • Cline, Roo Code and Continue compared: agent loop, modes, autocomplete, extensibility
    • How they differ from Claude Code, GitHub Copilot and Cursor
    • Licence, community and maintenance activity as selection criteria
  2. Module 2

    Bring your own model

    • Provider setup: Anthropic, OpenAI, Mistral, Google, OpenRouter and OpenAI-compatible endpoints
    • Matching models to planning, editing and autocomplete
    • Exercise: one ticket, two models, results and costs compared
  3. Module 3

    Day-to-day agent work

    • Plan and act modes, custom modes and role-specific instructions
    • Approvals for file edits and terminal commands, checkpoints and rollback
    • Rules files and project context that encode your team conventions
  4. Module 4

    Cost transparency

    • Reading token and cost displays per task correctly
    • Context hygiene: what makes a task expensive and how to counter it
    • Team budgets and cheaper models for routine work
  5. Module 5

    The enterprise model gateway

    • Gateway pattern: one endpoint, virtual keys, budgets, logging
    • Options such as LiteLLM or your cloud provider's AI platform
    • Exercise: switch all three agents to the company gateway

Why open-source agents in VS Code?

Commercial coding tools bundle editor, agent and model into one subscription. Open-source agents separate these layers: you keep VS Code, choose the extension and decide which model processes your code under which contract.

That freedom comes with responsibility: someone has to select models, manage keys, watch costs and keep configurations consistent. We show who owns which part.

The gateway as your control point

Personal API keys for every developer quickly become unmanageable. A central gateway lets you approve models, apply data protection settings once, attribute costs to teams and switch providers without touching each workstation.

  • Only approved models and regions are reachable
  • Keys per team or project, revocable at any time
  • Budgets and alerts instead of surprises on the invoice
  • Logs for security reviews, agreed with your works council and data protection officer

How the day runs

Participants work in their own VS Code installations on your repositories, or on a sandbox repository if code must not leave your network. Our AI Code Engineers bring prepared gateway configurations. Remote or on-site, in English or German.

Frequently asked questions

Are Cline, Roo Code and Continue free?

The extensions are open source, but the models are not: you pay your model provider or run models yourself. That is exactly why cost transparency and the gateway are core parts of the workshop.

Can we use Claude models with these agents?

Yes. All three can connect to Claude, either directly through the Anthropic API or through Amazon Bedrock and Google Vertex AI if traffic should stay within your existing cloud contracts.

Which of the three should we standardise on?

That depends on how your team works: Cline and Roo Code are built for autonomous agent tasks, while Continue is also strong at autocomplete and runs in JetBrains IDEs. Many teams allow more than one tool behind the same gateway.

Workshop

Get in touch

Want to run open-source coding agents with full control over models and costs?

Tell us roughly how many people will join and pick dates that work – we will get back to you within 24 hours with a proposal for content, format and schedule.

Prefer to talk first?

15 minutes, free and without obligation – with our AI Trainers and AI Code Engineers.

Book a call for free (opens in new tab) +49 761 519 751 50 (24/7)

Request a training / workshop

Workshop: Cline, Roo Code & Continue: Open-Source Coding Agents Workshop · 1 day

Dates that work for you (optional)

Suggestions about 2 weeks ahead – pick as many as you like

Reply usually within 24 hours · jocapps is ISO 27001 certified · Privacy