FAQ
The questions that come up before signup. If yours isn't here, email support@flarecode.sh — we'll answer and add it.
FlareCode is a cloud control tower for a fleet of coding agents. Connect a repo, describe a task in plain English, and an agent works in an isolated sandbox and opens a Pull Request you review. Run it across as many repos as you have.
Builders who don't sit still — solo founders, indie hackers, and small teams running several repos at once. If you've ever wanted to ship a fix from your phone, this is for you.
Those are inline-completion editors. You drive, they assist. FlareCode is the opposite: you describe the outcome, walk away, and review the PR. Different tool for a different moment.
Devin is a strong autonomous engineer aimed at teams and larger codebases. FlareCode is built for one person running a portfolio of repos: a fleet rail across every repo, your choice of model (Kimi, Claude, GPT, Gemini, or your own key), a per-task spend limit you set, and inference billed at provider cost. Pick Devin for team-scale autonomy; pick FlareCode to run many repos yourself.
No. Agent work always lands on a branch named flarecode/task-<id> and opens a Pull Request. Branch protection rules will reject direct pushes to default; the agent enforces this client-side too.
Minimum scopes: contents:write, pull_requests:write, metadata:read, actions:read. We do not request admin:repo_hook, admin:org, or delete:repo. You install per-repo, not per-org.
Each agent runs in an isolated sandbox on our infrastructure. We retain a private, encrypted workspace snapshot for the life of the agent so work can resume after idle pauses or restarts. Snapshots are deleted when you destroy the agent or account, and we never train on your code.
Yes. Anthropic, OpenAI, Google, OpenRouter, or a custom OpenAI-compatible provider. The BYOK plan exists exactly for this. You can also mix BYOK and bundled inference per agent on the Fleet plan.
Every task runs to a plan and stops at a per-task spend limit you set. If it stalls or hits the limit, we stop the task, refund the platform fee for that task, and post the log. Retries on the same task are free.
Secrets in your repo's .env are never read into the agent's working memory or model context. We sanitize log streams for known secret patterns (API keys, tokens, cloud credentials). No secret material persists in our logs.
Auto-merge is opt-in on the Pro plan only. By default every change gates on a human review. We will not auto-merge without explicit consent — pick the repos and the conditions yourself.
TypeScript, JavaScript, Python, Go, Rust, Ruby — anything the underlying coding agent handles well. The sandbox image ships with most common toolchains preinstalled.
Not today. The mobile-async surface is the web app, GitHub mobile, Slack, and email. Native iOS and macOS apps are roadmap items gated on customer demand.
Yes. FlareCode is in open beta. Sign in with GitHub and start on the free plan.
Get started
free to start · usage billed at cost · cancel anytime