# Multi-file edits in an existing repository

How to choose tools for changes that touch many files in a codebase you already have, and how to keep those changes reviewable. The groups separate in-editor agents, terminal agents, cloud agents and review tools, and this index does not rank them.

Canonical URL: https://globalaiproductindex.com/guides/existing-repo-multi-file-edits/

## How to choose

**Who this is for.** Developers and teams who need to change many files at once in a repository that already exists — refactors, migrations, renames, dependency upgrades or framework changes.

**Who this is not for.** Greenfield prototyping where a whole app is generated from a prompt, or teams that only need chat and autocomplete (see the AI coding tools guide).

- Decide where the change should run: inside an editor you keep open, in your terminal against your working copy, or in a cloud sandbox that opens a pull request.
- Check how the tool builds context for a large repository — repository indexing versus reading files per task — and whether that fits private code or monorepos.
- Confirm the review step: can you inspect and edit the multi-file diff before it is applied or committed?
- Prefer tools that can run your tests and linters, so multi-file changes are validated rather than assumed correct.
- Check data-retention and licensing terms before pointing an agent at private code.
- Write the task boundary down before the agent starts: the outcome you expect, the files it may touch, and the command that proves the change is complete.
- Create a rollback point first — a clean branch or a separate git worktree — so an unwanted multi-file change can be undone in one step.
- Keep the diff small enough to review in one sitting, and review at file level rather than accepting the whole change at once.

## Agentic editors that work across a whole repository

Editors where an agent can edit many files at once, run project tasks, and show a reviewable diff before you accept it. Unlike the terminal group next, the work stays inside an editor session rather than a shell you already script.

- [Cursor](https://globalaiproductindex.com/products/cursor/) (Anysphere) — Coding & Developer Tools, Freemium, launched 2023-03. AI-powered code editor that provides autocompletion, chat, and codebase-aware edits for software development.
- [Windsurf](https://globalaiproductindex.com/products/windsurf/) (Cognition) — Coding & Developer Tools, Freemium, launched 2024-11. Agentic AI code editor, formerly Codeium, that Cognition now ships as Devin Desktop, the IDE for its local and cloud agents.
- [Kiro](https://globalaiproductindex.com/products/kiro/) (Amazon Web Services) — Coding & Developer Tools, Freemium, launched 2025-07. Agentic IDE from AWS that turns prompts into executable specs and coordinates AI agents from prototype to production.
- [Google Antigravity](https://globalaiproductindex.com/products/antigravity/) (Google) — Coding & Developer Tools, Freemium, launched 2025-11. Agent-first development environment from Google where Gemini-powered agents plan, write, and verify code across editor, terminal, and browser.
- [Qoder](https://globalaiproductindex.com/products/qoder/) (Alibaba) — Coding & Developer Tools, Freemium, launched 2025-08. Agentic coding platform from Alibaba with an AI IDE and CLI that handle repository-wide context, delegated tasks, and code review.

## Terminal and CLI agents

Agents that run in a terminal, can use the shell, tests and git, and iterate until the build or the tests pass. Unlike the cloud group below, they run where you already have the repo, not in a vendor-hosted environment that opens a pull request later.

- [Claude Code](https://globalaiproductindex.com/products/claude-code/) (Anthropic) — Coding & Developer Tools, Paid, launched 2025-02. Anthropic's agentic coding tool that works in the terminal, editing files, running commands, and completing tasks across a codebase.
- [Codex](https://globalaiproductindex.com/products/openai-codex/) (OpenAI) — Coding & Developer Tools, Paid, launched 2025-05. OpenAI's software engineering agent that writes features, fixes bugs, and proposes pull requests from cloud sandboxes or the terminal.
- [Gemini CLI](https://globalaiproductindex.com/products/gemini-cli/) (Google) — Coding & Developer Tools, Open (open weights/source), launched 2025-06. Google's open-source terminal agent that brings Gemini models to the command line for coding, research, and task automation.
- [Aider](https://globalaiproductindex.com/products/aider/) (Aider AI LLC) — Coding & Developer Tools, Open (open weights/source), launched 2023. Open-source command-line tool for AI pair programming that edits code in your local git repository.
- [Cline](https://globalaiproductindex.com/products/cline/) (Cline) — Coding & Developer Tools, Open (open weights/source), launched 2024. Open-source autonomous coding agent for editors and terminals that can edit files, run commands, and use the browser with your approval.
- [Amp](https://globalaiproductindex.com/products/amp/) (Sourcegraph) — Coding & Developer Tools, Paid, launched 2025-07. Agentic coding tool from Sourcegraph that runs locally or in remote orbs, writing and shipping code from the terminal, editors, web, and phone.
- [goose](https://globalaiproductindex.com/products/goose/) (Block / Agentic AI Foundation) — Agents & Automation, Open (open weights/source), launched 2025-01. Open-source, extensible AI agent from Block that runs on your machine as a desktop app, CLI, and API, using any LLM and MCP extensions.

## Cloud and long-running agents

Agents that run tasks in a cloud environment, often opening a pull request you review later. Unlike the review group next, they produce the change; the review tools only inspect a diff that already exists.

- [Devin](https://globalaiproductindex.com/products/devin/) (Cognition AI) — Coding & Developer Tools, Paid, launched 2024-03. Autonomous AI software engineer that plans and completes end-to-end development tasks in its own sandboxed workspace.
- [Jules](https://globalaiproductindex.com/products/jules/) (Google) — Coding & Developer Tools, Freemium, launched 2025-05. Google's asynchronous coding agent that clones a repository, plans changes, and completes tasks like bug fixes in the background.
- [OpenHands](https://globalaiproductindex.com/products/openhands/) (All Hands AI) — Coding & Developer Tools, Open (open weights/source), launched 2024-03. Open-source platform for AI software development agents that can modify code, run commands, and browse the web (formerly OpenDevin).
- [Factory](https://globalaiproductindex.com/products/factory/) (Factory AI) — Coding & Developer Tools, Paid, launched 2023. Agent-native software development platform from Factory AI that uses Droids to plan, write, test, and ship code across the software lifecycle.

## Review and guardrails before merge

Tools that review the resulting diff, including AI-generated code, for quality, security and consistency. Unlike the editor group above, they do not make the multi-file edit; they are the gate before merge.

- [CodeRabbit](https://globalaiproductindex.com/products/coderabbit/) (CodeRabbit) — Coding & Developer Tools, Freemium, launched 2023. AI code-review assistant that reviews pull requests and suggests changes.
- [Qodo](https://globalaiproductindex.com/products/qodo/) (Qodo) — Coding & Developer Tools, Freemium, launched 2023. Code integrity platform (formerly CodiumAI) for AI-generated tests, reviews, and code suggestions.
- [SonarSource](https://globalaiproductindex.com/products/sonarsource/) (SonarSource Sàrl) — Coding & Developer Tools, Freemium, launched 2008. Code quality and security platform (SonarQube) that analyzes source code, including AI-generated code, with self-hosted, cloud, and IDE editions.
- [Codacy](https://globalaiproductindex.com/products/codacy/) (Qamine Portugal S.A.) — Coding & Developer Tools, Freemium, launched 2013. Automated code review and code quality platform that scans repositories and AI-generated code for quality, security, and coverage issues, with AI features for pull requests and coding agents.

## What was actually run

Executed on 2026-10-09 on one machine (macOS, arm64). The run record is linked in full below.

The runner created the scratch repository described in the fixture, asked three small local models through Ollama to fix it, applied whatever they returned and re-ran the tests. All three failed. The hosted coding agents on this page were not executed.

### Method

- Fixture: a real git repository with src/calc.js (wrong divisor), src/format.js (missing empty-list branch) and a fixed test file. node --test reports 2 pass / 3 fail before any model runs, and the tests may not be changed.
- Prompt: one request per model — temperature 0, seed 42, num_predict 1200. It asks for the failing repository to be fixed and for each changed file to be returned in full in a fixed FILE marker format. The exact template is in scripts/run-t17-coding.mjs.
- The runner writes only files under src/, then runs node --test and records the raw output, the git diff --stat and whether the suite passes. Nothing is retried or edited.
- Every model received the identical fixture and prompt; only the model name differed between the three attempts.

### Run records

| Run (UTC) | Runtime | Models | Result | Raw record |
| --- | --- | --- | --- | --- |
| 2026-10-09T12:13Z | Ollama — Runtime version not stored in this record. | qwen2.5-0.5b · SmolLM2-360M · TinyLlama-1.1B | Baseline 2 pass / 3 fail. All three models failed the task: qwen2.5-0.5b rewrote two src/ files with truncated content (14.2 s), SmolLM2-360M and TinyLlama-1.1B returned prose walkthroughs instead of files, so nothing could be applied (21.8 s and 8.8 s). The recorded per-model times sum to 44.8 s. | [docs/tests/existing-repo-coding/runs/2026-10-09T12-13-58-519Z.json](https://globalaiproductindex.com/tests/existing-repo-coding/runs/2026-10-09T12-13-58-519Z.json) |

### What the run shows

- The fixture is a real, checkable task with a fixed baseline and is not passable by accident: the fix needs an edit in calc.js and a branch in format.js.
- The runner can only write files under src/, so a model cannot pass by rewriting the tests — the guard was not needed in this run, but it is part of the method.
- At this parameter count the three local models are not a substitute for the hosted coding agents indexed on this page. That is a finding about these three models, not a quality judgement on any product.

### Failures that were kept

- qwen2.5-0.5b followed the output format but emitted the prompt's own placeholder text as file content: the applied src/calc.js begins with the literal text of the placeholder token, so node --test died with SyntaxError: Unexpected token '<'.
- SmolLM2-360M answered with a numbered “open your text editor” walkthrough, so the parser found no FILE markers and wrote no file; the suite stayed at its 2 pass / 3 fail baseline.
- TinyLlama-1.1B returned a shorter prose walkthrough with the same outcome: nothing applied, 2 pass / 3 fail. None of the three fixed the bug.

### What this does not show

- Three small local models, one three-file repository, one attempt each, one machine. Cursor, Claude Code, GitHub Copilot, Windsurf, Replit Agent, Aider, Cline and Tabnine were not executed.
- Two earlier records from the same day (runs/2026-10-09T12-12-26-217Z.json and runs/2026-10-09T12-12-52-783Z.json) are kept unedited: in them the harness's baseline command was wrong (Cannot find module …/test), so the fixture counts read “?” and 1/0. They are calibration attempts, not results.
- The record stores the raw output, the diff stat and the first lines of the test output, not a token-level log; wall-clock is specific to this machine.
- No score and no ranking is derived from one repository and one attempt per model.

**Cost.** No account, no API key, no paid plan — the models ran locally through Ollama. The measured cost was electricity for running the machine; no metered amount was recorded, so none is published.

### Method, fixture and raw outputs

Each label is the file's path in the source repository; the link opens a byte-identical copy of it published on this site (the first entry opens the index of all copies), so the record can be read without repository access.

- [docs/tests/](https://globalaiproductindex.com/tests/index.md)
- [docs/tests/existing-repo-coding/fixture.md](https://globalaiproductindex.com/tests/existing-repo-coding/fixture.md)
- [scripts/run-t17-coding.mjs](https://globalaiproductindex.com/tests/scripts/run-t17-coding.mjs)
- [docs/tests/existing-repo-coding/runs/2026-10-09T12-13-58-519Z.json](https://globalaiproductindex.com/tests/existing-repo-coding/runs/2026-10-09T12-13-58-519Z.json)
- [docs/tests/existing-repo-coding/runs/2026-10-09T12-12-26-217Z.json](https://globalaiproductindex.com/tests/existing-repo-coding/runs/2026-10-09T12-12-26-217Z.json)
- [docs/tests/existing-repo-coding/runs/2026-10-09T12-12-52-783Z.json](https://globalaiproductindex.com/tests/existing-repo-coding/runs/2026-10-09T12-12-52-783Z.json)

### What was not executed

[Cursor](https://globalaiproductindex.com/products/cursor/) · [Claude Code](https://globalaiproductindex.com/products/claude-code/) · [GitHub Copilot](https://globalaiproductindex.com/products/github-copilot/) · [Windsurf](https://globalaiproductindex.com/products/windsurf/) · [Replit](https://globalaiproductindex.com/products/replit/) · [Aider](https://globalaiproductindex.com/products/aider/) · [Cline](https://globalaiproductindex.com/products/cline/) · [Tabnine](https://globalaiproductindex.com/products/tabnine/)

This round did not execute these applications, so no measured result is reported for any of them. Some of them need an account or a particular plan before they can be run: Claude Code needs an available access authorisation, and the documented paths are a Claude subscription (Pro, Max, Team or Enterprise), a Claude Console account, or access through a supported cloud provider — so a paid Claude subscription is not the only route — while the hosted agents need a signed-in account. That is not true of every product listed — Aider and Cline are open source and can connect to a local model, so neither needs a paid plan. No further reason was recorded for this round, so where a condition was not confirmed it is stated as unconfirmed rather than assumed. Their pages and comparisons report documented behaviour with a source and a check date.


## FAQ

**Does this guide rank multi-file tools?**

No. It groups indexed products by how they handle changes across an existing repository. It assigns no scores and reports no benchmark results.

**Will an agent understand a large codebase?**

Capability varies: some tools index the repository, others read files per task. Check each product's documentation for how it builds context and what context limits apply.

**What is the smallest safe way to start?**

Start in a read-only or suggestion-only mode on a branch, review the diff, and only then enable edits and command execution. Keep changes small enough to review in one sitting.

**Is a cloud agent better than a local one?**

It depends on constraints. Cloud agents can run long tasks without your machine, while local tools keep code in your environment. Both are listed here without preference.

**How do I stop an agent from rewriting the whole repository?**

Constrain the task before it runs: name the files or directories in scope, state what must not change, and ask for the smallest diff that achieves the outcome. Then review the diff before applying it. Published multi-agent workflows use the same idea — an explicit task boundary plus an isolated branch or worktree. This guide reports no measured results.

## Sources and record dates

Record dates describe index updates, not independent product tests. Check linked documentation for current features and pricing.

### Cursor

Record updated: 2026-10-09

- [Official website](https://cursor.com)
- [Documentation](https://cursor.com/docs)
- [Official pricing](https://cursor.com/pricing)
- [AI Coding Agent for Building Ambitious Software | Cursor](https://cursor.com/)
- [Pricing](https://cursor.com/pricing)
- [Documentation](https://cursor.com/docs)

### Windsurf

Record updated: 2026-10-10

- [Devin Desktop (windsurf.com now redirects here)](https://windsurf.com)
- [Devin Desktop documentation](https://docs.devin.ai/desktop/getting-started)
- [Devin pricing](https://devin.ai/pricing)
- [Pricing](https://devin.ai/pricing)
- [Documentation](https://docs.devin.ai/desktop/getting-started)

### Kiro

Record updated: 2026-09-10

- [Official website](https://kiro.dev)
- [InfoQ: Amazon introduces Kiro, the spec-driven agentic AI IDE](https://www.infoq.com/news/2025/08/aws-kiro-spec-driven-agent/)
- [Documentation](https://kiro.dev/docs/)
- [Official pricing](https://kiro.dev/pricing/)
- [Pricing](https://kiro.dev/pricing/)
- [Documentation](https://kiro.dev/docs/)

### Google Antigravity

Record updated: 2026-10-09

- [Official website](https://antigravity.google)
- [Ars Technica: Google unveils Gemini 3 AI model and AI-first IDE called Antigravity](https://arstechnica.com/google/2025/11/google-unveils-gemini-3-ai-model-and-ai-first-ide-called-antigravity/)
- [Google Antigravity](https://antigravity.google/)
- [Documentation](https://antigravity.google/docs/getting-started)
- [Official pricing](https://antigravity.google/pricing)
- [Pricing](https://antigravity.google/pricing)
- [Documentation](https://antigravity.google/docs/getting-started)

### Qoder

Record updated: 2026-10-10

- [Official website](https://qoder.com)
- [Yahoo Finance: Alibaba launches Qoder, an agentic coding platform for real software](https://finance.yahoo.com/news/alibaba-launches-qoder-agentic-coding-133000732.html)
- [Official pricing](https://qoder.com/pricing)
- [Pricing](https://qoder.com/pricing)

### Claude Code

Record updated: 2026-10-09

- [Official website](https://www.anthropic.com/claude-code)
- [Documentation](https://code.claude.com/docs/en/overview)
- [Official pricing](https://claude.com/pricing)
- [Claude Code billing and usage](https://code.claude.com/docs/en/costs)
- [Pricing](https://claude.com/pricing)
- [Documentation](https://code.claude.com/docs/en/overview)

### Codex

Record updated: 2026-09-04

- [Official website](https://openai.com/codex)
- [OpenAI Codex (Wikipedia)](https://en.wikipedia.org/wiki/OpenAI_Codex)

### Gemini CLI

Record updated: 2026-09-10

- [Source code (GitHub)](https://github.com/google-gemini/gemini-cli)
- [Google announcement](https://blog.google/innovation-and-ai/technology/developers-tools/introducing-gemini-cli-open-source-ai-agent/)

### Aider

Record updated: 2026-10-10

- [Official website](https://aider.chat)
- [Source code (GitHub)](https://github.com/Aider-AI/aider)
- [Documentation](https://aider.chat/docs/)
- [Documentation](https://aider.chat/docs/)

### Cline

Record updated: 2026-10-10

- [Official website](https://cline.bot)
- [Source code (GitHub)](https://github.com/cline/cline)
- [Cline - AI Coding, Open Source and Open Choice](https://cline.bot/)
- [Documentation](https://docs.cline.bot/cline-overview)
- [Official pricing](https://cline.bot/pricing)
- [Pricing](https://cline.bot/pricing)
- [Documentation](https://docs.cline.bot/cline-overview)

### Amp

Record updated: 2026-10-09

- [Official website](https://ampcode.com/)
- [Amp pricing](https://ampcode.com/pricing)
- [Changes to Cody Free, Pro, and Enterprise Starter plans](https://sourcegraph.com/blog/changes-to-cody-free-pro-and-enterprise-starter-plans)
- [Introduction | Amp Docs](https://ampcode.com/docs)
- [Pricing](https://ampcode.com/pricing)
- [Documentation](https://ampcode.com/docs)

### goose

Record updated: 2026-10-09

- [Official website](https://goose-docs.ai/)
- [GitHub repository](https://github.com/aaif-goose/goose)
- [Quickstart | goose | Your open source AI agent](https://goose-docs.ai/docs/quickstart/)
- [Documentation](https://goose-docs.ai/docs/quickstart)

### Devin

Record updated: 2026-10-10

- [Official website](https://devin.ai)
- [Cognition AI](https://cognition.ai)
- [Devin: the autonomous software engineer](https://devin.ai/)
- [Official pricing](https://devin.ai/pricing)
- [Documentation](https://docs.devin.ai)
- [Pricing](https://devin.ai/pricing)
- [Documentation](https://docs.devin.ai)

### Jules

Record updated: 2026-09-10

- [Official website](https://jules.google)
- [Google announcement](https://blog.google/innovation-and-ai/models-and-research/google-labs/jules-now-available/)

### OpenHands

Record updated: 2026-10-10

- [Official website](https://www.all-hands.dev)
- [Source code (GitHub)](https://github.com/All-Hands-AI/OpenHands)

### Factory

Record updated: 2026-09-04

- [Official website](https://factory.ai)
- [Factory 2.0: From coding agents to software factories](https://factory.ai/news/software-factory)
- [Pricing](https://factory.ai/pricing)
- [Documentation](https://docs.factory.ai/)

### CodeRabbit

Record updated: 2026-09-10

- [Official website](https://coderabbit.ai)
- [Documentation](https://docs.coderabbit.ai)
- [Official pricing](https://www.coderabbit.ai/pricing)
- [Pricing](https://www.coderabbit.ai/pricing)
- [Documentation](https://docs.coderabbit.ai)

### Qodo

Record updated: 2026-10-10

- [Official website](https://www.qodo.ai)
- [Documentation](https://docs.qodo.ai)
- [Official pricing](https://www.qodo.ai/pricing/)
- [Pricing](https://www.qodo.ai/pricing/)
- [Documentation](https://docs.qodo.ai)

### SonarSource

Record updated: 2026-09-10

- [Official website](https://www.sonarsource.com)
- [Documentation](https://docs.sonarsource.com)
- [SonarSource (Wikipedia)](https://en.wikipedia.org/wiki/SonarSource)
- [Pricing](https://www.sonarsource.com/plans-and-pricing/)
- [Documentation](https://docs.sonarsource.com)

### Codacy

Record updated: 2026-09-10

- [About us](https://www.codacy.com/about)
- [Terms of Service](https://www.codacy.com/terms)
- [Blog: Open for business (February 2014)](https://web.archive.org/web/20150228050524/http://blog.codacy.com/2014/02/06/open-for-business/)
- [Pricing](https://www.codacy.com/pricing)
- [Documentation](https://docs.codacy.com/)

---
Data from the [Global AI Product Index](https://globalaiproductindex.com/), licensed CC BY 4.0.