Chapter C73 · Interactive Course
Other coding agents and interfaces
Terminal, editor and cloud examples, open clients with configured providers, explicit file context and frameworks.
Core Learning Concept
Place the product cards by their interface, model-provider arrangement, and execution model. Distinguish a ready-to-use agent product from a framework for building one. The exercise teaches which questions to ask; actual availability, settings, and requirements should be checked in each product’s current documentation.
What You’ll Learn in This Chapter
-
Gemini CLI brings an agent to the terminal
Recognize a terminal agent using Gemini models and project context.
Gemini CLI provides an agent interface inside the terminal. It interacts with local files under configured directory permissions while models process remotely. Always verify your current folder path and active permissions before execution.
-
Cursor brings agent work into an editor
Identify a file-centered editor workflow.
Cursor integrates AI assistance and agent workflows into an editor. That can place file navigation and diffs near the conversation. Check the selected mode, included context, execution controls, and actual changes. An editor integration helps the workflow but does not remove the need for task-specific verification.
-
Copilot has different assistance modes
Distinguish a completion suggestion from a delegated cloud task.
GitHub Copilot offers several modes and surfaces with different capabilities. A code suggestion, local agent session, and cloud agent task are not interchangeable. Identify the specific mode and what it can access, execute, and return. Compare outcomes using the intended workflow rather than the product label alone.
-
OpenCode separates client from provider
Recognize an open-source agent connected to configured inference.
OpenCode cleanly separates the client interface from the backend model provider. Switching providers changes latency, cost, and capabilities while preserving familiar controls. An open-source client does not mean the model weights or execution run locally.
-
Aider supports terminal pair programming
Identify explicit file context in a small editing workflow.
Aider offers terminal pair programming centered around explicit file selection and Git commits. Shaping which files enter the context window keeps edits focused. Always inspect diffs and git log history after every change.
-
Frameworks help developers assemble agents
Distinguish a library for building orchestration from a ready-to-use app.
An agent framework supplies building blocks for developers to assemble model calls, tools, state, and orchestration. It is different from a fully configured end-user agent product. The developer must still design permissions, error handling, evaluation, and deployment. A framework makes construction possible, not automatically reliable.
-
Place the product cards
Sort the named examples by interface, execution location and model relationship, without producing a universal ranking.