Using RepoWise with Cloud AI Tools

Your AI coding tools work great on your local machine. But what happens when Devin, Jules, or Copilot Cloud Agent pick up a task? They clone your repo from Git into a cloud VM, and they can only see what's committed.

RepoWise generates context files locally by default. For cloud-based AI tools, you need to commit those files to your repository so the cloud agent finds them when it clones.

This guide walks you through the setup for each cloud tool we support.

How Cloud Tools Differ from Local Tools

Local tools like Cursor, Claude Code CLI, or Cline run on your machine. They read context files directly from your filesystem. No Git commit needed. RepoWise handles this automatically.

Cloud tools run in remote environments. When you assign a task to Devin or kick off a Cursor Background Agent, the tool spins up a VM, clones your repo from GitHub (or GitLab/Bitbucket), and starts working. If your context files aren't committed, the cloud agent starts blind.

The fix is simple: commit your repowise-context/ folder and tool config files to Git.

Prerequisites

Before setting up a cloud tool, make sure you've completed the standard RepoWise onboarding:

1. Account created at app.repowise.ai
2. CLI installed and repowise create run in your project
3. Context files generated in repowise-context/
4. Tool config files created (e.g., AGENTS.md, CLAUDE.md)

New to RepoWise?

Complete the Quick Start first. Come back here once your context files are generated.

Choose Your Cloud Tool

Select your cloud tool below for a complete setup guide.

Cursor - Background Agent

What is Cursor's Background Agent?

Cursor's Background Agent runs coding tasks in an isolated AWS virtual machine. It clones your repo from GitHub or GitLab, works on the task autonomously, and opens a pull request when done. You can run multiple agents in parallel without tying up your local editor.

Files the Background Agent needs

File	Purpose
repowise-context/	Your AI-optimized context files
.cursor/rules/repowise.mdc	Cursor-specific rules pointing to context

Step 1: Run RepoWise onboarding

If you haven't already, navigate to your project and generate context:

cd your-project
repowise create

Select Cursor when prompted for AI tools.

Step 2: Remove repowise-context from .gitignore

RepoWise adds repowise-context/ to .gitignore by default (for local-only setups). Remove it:

# Open .gitignore and delete the line containing "repowise-context"
sed -i '' '/repowise-context/d' .gitignore

Verify it's removed:

grep "repowise-context" .gitignore
# Should return nothing

Step 3: Commit context files

git add repowise-context/
git commit -m "chore: add RepoWise context for cloud AI tools"

Step 4: Commit Cursor rules

git add .cursor/rules/repowise.mdc
git commit -m "chore: add Cursor rules for RepoWise context"

Step 5: Push to remote

git push

Step 6: Verify it works

1. Open Cursor and start a Background Agent task on your repo
2. Once the agent starts, check its output. It should reference your project architecture and patterns from the context files
3. The agent reads .cursor/rules/repowise.mdc automatically, which points it to repowise-context/

Step 7: Keep context updated

After each RepoWise sync (triggered automatically when you push code), commit the updated files:

git add repowise-context/ .cursor/rules/repowise.mdc
git commit -m "chore: update RepoWise context"
git push

tip

Run repowise status to check when your last sync completed.

Learn more: Cursor Cloud Agent docs

Claude Code - Web & GitHub Action

What are Claude Code's cloud modes?

Claude Code has two cloud modes:

• claude.ai/code - A web-based agent that clones your repo into a cloud sandbox, makes changes, and creates pull requests.
• GitHub Action - Runs Claude Code in your CI pipeline. Responds to @claude mentions in issues and PRs, or runs automated code reviews.

Both modes clone your repo and read CLAUDE.md automatically.

Files Claude Code needs

File	Purpose
repowise-context/	Your AI-optimized context files
CLAUDE.md	Claude Code instructions pointing to context

Step 1: Run RepoWise onboarding

cd your-project
repowise create

Select Claude Code when prompted for AI tools.

Step 2: Remove repowise-context from .gitignore

sed -i '' '/repowise-context/d' .gitignore

Step 3: Commit context files

git add repowise-context/
git commit -m "chore: add RepoWise context for cloud AI tools"

Step 4: Commit CLAUDE.md

git add CLAUDE.md
git commit -m "chore: add CLAUDE.md with RepoWise context references"

note

If you already have a CLAUDE.md with custom instructions, RepoWise adds its content between <!-- repowise-start --> and <!-- repowise-end --> markers. Your existing content is preserved.

Step 5: Push to remote

git push

Step 6: Verify it works

For claude.ai/code:

1. Go to claude.ai/code
2. Connect your GitHub account and select your repo
3. Start a task. Claude will read CLAUDE.md and find repowise-context/.
4. Ask "What is the project architecture?" It should answer using your context files.

For GitHub Action:

1. Set up anthropics/claude-code-action in your repo (setup guide)
2. Comment @claude what is the project architecture? on any PR
3. Claude reads CLAUDE.md from the checked-out repo and responds with context-aware answers

Step 7: Keep context updated

git add repowise-context/ CLAUDE.md
git commit -m "chore: update RepoWise context"
git push

Learn more: Claude Code GitHub Actions docs | Claude Code on the web

GitHub Copilot - Cloud Agent

What is Copilot's Cloud Agent?

GitHub's Copilot Cloud Agent is an autonomous coding agent that runs in a GitHub Actions-powered environment. You assign it a GitHub issue, and it clones your repo, analyzes the codebase, implements changes, and opens a pull request, all without your local machine.

Files the Cloud Agent needs

File	Purpose
repowise-context/	Your AI-optimized context files
.github/copilot-instructions.md	Copilot-specific instructions pointing to context
AGENTS.md	Cross-tool instructions (also read by Copilot)

Step 1: Run RepoWise onboarding

cd your-project
repowise create

Select GitHub Copilot when prompted for AI tools.

Step 2: Remove repowise-context from .gitignore

sed -i '' '/repowise-context/d' .gitignore

Step 3: Commit context files

git add repowise-context/
git commit -m "chore: add RepoWise context for cloud AI tools"

Step 4: Commit Copilot instructions and AGENTS.md

git add .github/copilot-instructions.md AGENTS.md
git commit -m "chore: add Copilot instructions and AGENTS.md for RepoWise"

Step 5: Push to remote

git push

Step 6: Verify it works

1. Go to a GitHub issue in your repo
2. Assign the issue to Copilot
3. Copilot spins up a cloud environment, clones your repo, and reads .github/copilot-instructions.md
4. Check the agent's plan. It should reference your project's architecture and patterns.

info

Copilot Cloud Agent is available to Copilot Enterprise and Copilot Pro+ customers.

Step 7: Keep context updated

git add repowise-context/ .github/copilot-instructions.md AGENTS.md
git commit -m "chore: update RepoWise context"
git push

Learn more: About Copilot Cloud Agent

Codex - ChatGPT Codex

What is ChatGPT Codex?

OpenAI's Codex is a cloud coding agent available in ChatGPT. For every task, it creates an isolated cloud container, clones your repo from GitHub, and gets to work. It reads AGENTS.md from the repo root for project-specific instructions.

note

This is the cloud product at chatgpt.com/codex, not the open-source Codex CLI. The CLI runs locally and is already supported by RepoWise without extra setup.

Files Codex needs

File	Purpose
repowise-context/	Your AI-optimized context files
AGENTS.md	Instructions pointing to context (Codex reads this automatically)

Step 1: Run RepoWise onboarding

cd your-project
repowise create

Select Codex when prompted for AI tools.

Step 2: Remove repowise-context from .gitignore

sed -i '' '/repowise-context/d' .gitignore

Step 3: Commit context and config files

git add repowise-context/ AGENTS.md
git commit -m "chore: add RepoWise context for cloud AI tools"
git push

Step 4: Verify it works

1. Go to chatgpt.com/codex
2. Connect your GitHub account if you haven't already
3. Select your repo and create a task
4. Codex clones your repo into a sandbox and reads AGENTS.md
5. Ask it about your architecture. It should use RepoWise context to answer accurately.

Step 5: Keep context updated

git add repowise-context/ AGENTS.md
git commit -m "chore: update RepoWise context"
git push

Learn more: Codex Cloud docs

Warp - Oz Cloud Agents

What are Warp's Cloud Agents?

Warp's Oz Cloud Agents run in Docker containers in the cloud. When triggered, they clone your configured GitHub repositories into the container, run setup commands, and execute coding tasks autonomously. You can run agents interactively or as background tasks.

Files Warp Cloud Agents need

File	Purpose
repowise-context/	Your AI-optimized context files
AGENTS.md	Instructions pointing to context

Step 1: Run RepoWise onboarding

cd your-project
repowise create

Select Warp when prompted for AI tools.

Step 2: Remove repowise-context from .gitignore

sed -i '' '/repowise-context/d' .gitignore

Step 3: Commit context and config files

git add repowise-context/ AGENTS.md
git commit -m "chore: add RepoWise context for cloud AI tools"
git push

Step 4: Verify it works

1. Create an Oz environment for your repo (quickstart)
2. Start a cloud agent task
3. The agent clones your repo and reads AGENTS.md automatically
4. Check that the agent references your project context in its responses

Step 5: Keep context updated

git add repowise-context/ AGENTS.md
git commit -m "chore: update RepoWise context"
git push

Learn more: Warp Cloud Agents docs

Jules - Google Jules

What is Google Jules?

Jules is Google's autonomous coding agent. It runs entirely in the cloud. There is no local mode. When you submit a task, Jules spins up a VM, clones your repo from GitHub, analyzes the codebase, and creates a pull request. It reads AGENTS.md for project-specific instructions.

Files Jules needs

File	Purpose
repowise-context/	Your AI-optimized context files
AGENTS.md	Instructions pointing to context

Step 1: Run RepoWise onboarding

cd your-project
repowise create

Select Google Jules when prompted for AI tools.

Step 2: Remove repowise-context from .gitignore

sed -i '' '/repowise-context/d' .gitignore

Step 3: Commit context and config files

git add repowise-context/ AGENTS.md
git commit -m "chore: add RepoWise context for cloud AI tools"
git push

Step 4: Verify it works

1. Go to jules.google and connect your GitHub account
2. Select your repo and submit a task (e.g., "Fix the bug described in issue #1")
3. Jules clones your repo, reads AGENTS.md, and finds repowise-context/
4. Check the agent's plan. It should reference your project's architecture.

info

Jules currently supports GitHub repositories only.

Step 5: Keep context updated

git add repowise-context/ AGENTS.md
git commit -m "chore: update RepoWise context"
git push

Learn more: Jules docs

Devin - by Cognition

What is Devin?

Devin is a cloud-only AI software engineer by Cognition. Each session runs in an isolated VM with a full Linux desktop, terminal, code editor, and browser. Devin clones your repo at the start of every session and reads multiple instruction files automatically, including AGENTS.md, CLAUDE.md, .cursorrules, and more.

Files Devin needs

File	Purpose
repowise-context/	Your AI-optimized context files
AGENTS.md	Instructions pointing to context (auto-read by Devin)

tip

Devin also auto-reads CLAUDE.md, .cursorrules, .windsurf, and .mdc files. If you've selected multiple tools in RepoWise, Devin benefits from all of them.

Step 1: Run RepoWise onboarding

cd your-project
repowise create

Select Devin when prompted for AI tools.

Step 2: Remove repowise-context from .gitignore

sed -i '' '/repowise-context/d' .gitignore

Step 3: Commit context and config files

git add repowise-context/ AGENTS.md
git commit -m "chore: add RepoWise context for cloud AI tools"
git push

Step 4: Verify it works

1. Open app.devin.ai and start a new session on your repo
2. Devin clones the repo and automatically ingests AGENTS.md
3. Ask Devin about your project architecture. It should answer using the RepoWise context.
4. Check the Knowledge section in Devin's UI to confirm AGENTS.md was loaded

Step 5: Keep context updated

git add repowise-context/ AGENTS.md
git commit -m "chore: update RepoWise context"
git push

note

Devin supports GitHub, GitLab, and Bitbucket repositories.

Learn more: Devin AGENTS.md docs | Devin Knowledge Onboarding

Junie - JetBrains CLI

What is Junie CLI?

Junie CLI is JetBrains' standalone coding agent that runs from the terminal, inside any IDE, or in CI/CD pipelines. When running in CI or triggered from GitHub/GitLab, Junie operates on a cloned copy of your repo and reads .junie/AGENTS.md for project guidelines.

Files Junie CLI needs

File	Purpose
repowise-context/	Your AI-optimized context files
.junie/AGENTS.md	Junie-specific guidelines pointing to context

Step 1: Run RepoWise onboarding

cd your-project
repowise create

Select JetBrains Junie when prompted for AI tools.

Step 2: Remove repowise-context from .gitignore

sed -i '' '/repowise-context/d' .gitignore

Step 3: Commit context and config files

git add repowise-context/ .junie/AGENTS.md
git commit -m "chore: add RepoWise context for cloud AI tools"
git push

Step 4: Verify it works

1. Run Junie CLI in your CI pipeline or from a GitHub/GitLab integration
2. Junie reads .junie/AGENTS.md and discovers repowise-context/
3. Tasks should reflect awareness of your project architecture and patterns

Step 5: Keep context updated

git add repowise-context/ .junie/AGENTS.md
git commit -m "chore: update RepoWise context"
git push

Learn more: Junie CLI announcement

Keeping Context Fresh

RepoWise automatically updates your context files when you push code. The background listener detects changes and refreshes repowise-context/ on your local machine.

For cloud tools, you need one extra step: commit and push the updated files.

After each sync, run:

git add repowise-context/ AGENTS.md  # or your tool's config file
git commit -m "chore: update RepoWise context"
git push

Automate it

Add a simple script to your workflow that commits context updates after each sync. You can check for changes with:

if [[ -n $(git status --porcelain repowise-context/) ]]; then
  git add repowise-context/ AGENTS.md
  git commit -m "chore: update RepoWise context"
  git push
fi

FAQ

My cloud tool doesn't see the context files. What's wrong?

Check three things: (1) repowise-context/ is not in .gitignore, (2) the folder is committed and pushed to your remote, and (3) the cloud tool is working on the correct branch.

My context is outdated in the cloud agent.

The cloud agent sees whatever is committed to Git. Run repowise status to check your last sync, then commit and push any updated files.

I don't see AGENTS.md in my repo.

Make sure you selected a tool that uses AGENTS.md during repowise create. You can re-run repowise sync to regenerate config files, or run repowise create again to re-select tools.

Can I use both local and cloud tools at the same time?

Absolutely. Local tools (like Cursor's editor mode or Claude Code CLI) continue reading from your filesystem as usual. Cloud tools read from the committed files in Git. Both can coexist. Just make sure you commit your context files for the cloud tools.

Will committing repowise-context/ clutter my repo?

The context files are structured Markdown, typically 10-30 files totaling a few hundred KB. They're lightweight and useful for any contributor (human or AI) who wants to understand your project.

What's Next

• Quick Start - set up RepoWise for the first time
• Connecting Repos - add more repositories
• Team Member Onboarding - invite your team

---

Found a mistake on this page? Let us know at support@repowise.ai.