Privacy Policy
Spark is local-first. Your agent, memory, missions, provider keys, and tools are meant to live on your machine, not in a Sparkswarm account. This policy explains what the website can see, what Spark keeps local, and what changes if you connect outside services.
What Spark keeps local
Spark stores its working state on your computer unless you explicitly connect a service that sends data elsewhere. That includes local memory, missions, logs, provider configuration, and generated tools. We do not receive a cloud copy of your local Spark workspace.
What the website collects
agent.sparkswarm.ai is a static website for reading docs and downloading installer files. Like most websites, our hosting provider may process basic request information such as IP address, browser, timestamp, requested URL, and error logs so the site can load and stay secure.
We do not use third-party ad tracking or session replay on this site. Theme preference is stored in your browser's local storage, not sent to us. If we add privacy-preserving analytics later, we will update this policy and the cookie page before relying on it.
Installer requests
When you download install.sh, install.ps1, checksums, or release manifests, the web server receives a normal file request. The installer itself runs locally and shows a dry-run plan before writing Spark files.
What happens when you connect third-party tools
Spark can use providers and tools you choose, such as Telegram, GitHub, Codex, Claude Code, OpenRouter, Ollama, LM Studio, or other LLM routes. When you authorize or configure a connection:
- Tokens and API keys are stored locally by Spark unless a specific integration says otherwise.
- The provider receives whatever prompts, messages, files, or metadata you send through that provider.
- Those providers have their own privacy policies and retention rules.
- Sparkswarm does not see your provider tokens or provider responses through the local installer flow.
Future swarm features
Spark Swarm is planned as an opt-in collective-learning layer. It is not required for local Spark. If we ship sharing features, they will be opt-in, scoped, and documented before data leaves your machine.
Children
Spark is a developer tool. It isn't designed for children under 13, and we don't knowingly collect information from them.
Your choices
Because Spark is local-first, most control happens on your machine: delete local files, rotate provider keys, disconnect integrations, or uninstall Spark. If you contact us by email and want that message deleted, write to [email protected].
Changes
If this policy changes materially, the date at the top will change. If a new feature sends Spark data to Sparkswarm, we will document that before enabling it.
Questions or privacy requests: [email protected].