Skip to content
ReviewRamble

The weekly brief

AI, reviewed, in your inbox.

  • New reviews of tools, models, MCP servers and repos, with the verdict up front
  • Re-checked prices and ratings when something changes
  • New analysis pieces as they are published

Models, agents, MCP servers and the tools actually worth your time.

One email a week. Unsubscribe with the link in any brief. How we use your email

Coding agent

A coding agent is an AI agent that works on software directly: it reads a codebase, edits files, runs commands and tests, and repeats until a change works, rather than only suggesting code for a person to paste.

Also called: AI coding agent

By Divya Sharma · Checked Nov 15, 2026 · How we review

Code completion suggests the next few lines. A coding agent takes a whole task, such as "fix this failing test" or "add a settings page", and carries it out across many files and steps.

How it works

The agent runs an agent loop with tools for reading and writing files, searching code and running shell commands. It can run the test suite, read the errors and try again. Some run in a terminal, such as Claude Code and Aider. Some live inside an AI code editor. Others run in the cloud on a copy of the repository and return a pull request, such as OpenAI Codex. Open-source options include OpenHands.

Results depend heavily on the agent harness: the tools, instructions and checks around the model. Project rule files and agent skills tell it how the codebase works.

Example

Given a bug report, an agent searches for the failing function, reproduces the bug with a test, edits the code, runs the tests and summarises the change.

Limits

Agents can produce code that passes tests but is wrong, rewrite more than asked, or run a harmful command. Review the diff, run the agent in an agent sandbox and limit what it can do without approval. Benchmarks such as SWE-bench give a rough idea of capability.

Related terms

AI code editor
An AI code editor is a code editor or IDE with language models built in, offering completions, chat about the codebase and an agent mode that can plan and make changes across many files.
Agent harness
An agent harness is the software around a model that turns it into a working agent: the loop, tools, system prompt, permissions, context management and any files used to carry progress between sessions.
Agent sandbox
An agent sandbox is an isolated environment, such as a container, virtual machine or operating-system restriction, in which an AI agent runs code and commands so it cannot reach files, networks or data outside set limits.
SWE-bench
SWE-bench is a coding benchmark built from real GitHub issues in popular Python projects; a model or agent must edit the repository so that the issue is fixed and the project's hidden tests pass.
Terminal-Bench
Terminal-Bench is a benchmark for AI agents that work in a command-line terminal, where each task gives the agent a real environment and is scored by tests that check whether the job was actually done.
Vibe coding
Vibe coding is building software by describing what you want to an AI model and accepting the code it writes without reading or reviewing it, judging the result only by whether the program seems to work.

Sources