Single API call to run a fully managed, sandboxed Claude Code, Codex or OpenCode. Use provider API keys or your Claude / ChatGPT subscription. Give your product SOTA coding ability in minutes
curl -X POST https://codecloud.dev/api/v1/agents \
-H "Authorization: Bearer $API_KEY" \
-d '{
"repo": "your-org/your-repo",
"agent_provider": "claude-code",
"agent_auth_method": "subscription",
"prompt": "Fix the failing tests in src/utils"
}'How it works
The same path whether you trigger it from your product, a ticket, or a cron job.
01
Install the GitHub app and choose which repositories the agent can touch.
02
One POST from your product, Linear, or any automation tool you already use.
03
The agent works in an isolated sandbox and opens a pull request when it’s done.
Runtimes
Official CLIs, fully managed. Authenticate with a provider API key or the subscription you already pay for.
claude-code
Anthropic’s official CLI. Use an API key, Amazon Bedrock, or connect Claude Pro, Max, Team, or Enterprise from the dashboard.
codex
OpenAI’s official CLI. Use an API key, or connect ChatGPT on the dashboard and we’ll restore the subscription for each run.
opencode
The default multi-model runtime. Bring your own Anthropic, OpenAI, Google, or other provider keys.
Security
Your code and credentials are protected at every step of a run.
API keys and subscription credentials protected with AES-256-GCM.
GitHub access uses encrypted, temporary credentials.
Each agent runs in its own secure environment.
Repository contents are deleted after every run.
Details in our Privacy Policy
Integrations
Drop coding agents into the stack you already have.
API
Start runs, poll status, receive webhooks. Add coding ability to any product or service.
View docsAutomation
Trigger agents from forms, email, Slack, or any other event — no extra infrastructure.
See n8n guideNative
Assign issues to @codecloud. Watch the agent work in comments, then talk back to course-correct.
CI
Auto-fix failing tests, generate docs on merge, or run review bots on every pull request.
View docsSchedules
Run agents daily, weekly, or on any cron. They work overnight and open PRs for the morning.
Go beyond Renovate — update packages and fix the breaking changes.
Dead code, linter debt, deprecated APIs — a few files at a time.
Regular scans, then patches for known vulnerabilities before they ship.
POST /api/v1/schedules
{
"type": "recurring",
"cron": "0 4 * * 1-5",
"repo": "your-org/your-repo",
"prompt": "Update npm dependencies and fix any breaking changes",
"auto_create_pr": true
}Capabilities
Custom tools, blank workspaces, and plan mode — without standing up your own runtime.
Pass HTTP endpoints so agents can call internal APIs, query data, or check flags while they work.
Tool use guideRun without a repo for data transforms, reports, or multi-system orchestration via tools.
Learn moreGenerate implementation plans grounded in the actual codebase before any code is written.
Plan mode guideAgents keep going until tests pass — not just until they think they’re finished.
How it loopsPricing
A fixed monthly price on top of your usual LLM usage — API keys or a Claude / ChatGPT subscription.
No plans available. Please configure plans in your Clerk dashboard.
Fully managed Claude Code, Codex, or OpenCode. API keys or your existing subscription. One API call.