Klaussy

← Back to Klaussy

Privacy Policy

Last updated: August 31, 2026 · Draft pending legal review.

1. What Klaussy collects

Klaussy is a desktop application. We do not run a backend, and we receive none of your data. Everything below leaves your machine only because you pointed Klaussy at a service and asked it to do something. Each provider's own policy governs what they keep.

  • Your AI coding agent. The agent you select receives the prompts you write and whatever repository context it reads to answer them. That agent may be Claude Code (Anthropic), Codex (OpenAI), Gemini or Antigravity (Google), Copilot (GitHub), Cursor (Anysphere), Cline, opencode, Kimi Code (Moonshot AI), or Aider. Which provider gets your code depends entirely on which agent you pick, per tab.
  • Your git forge. Cloning, browsing pull or merge requests, reading CI logs, and posting review comments send requests to GitHub, GitLab, or Bitbucket, through their CLIs or REST APIs using credentials you supply.
  • Discord or Slack, if you turn on session mirroring. Mirroring streams that session's agent output into a thread, and sends your replies back to the agent. Whatever the agent prints goes with it, which in practice includes file paths, diffs, and source code. This is off unless you enable it per session.
  • MCP servers you enable. Klaussy ships a catalog covering Asana, Atlassian (Jira and Confluence), Datadog, Figma, GitHub, Grafana, Honeycomb, Linear, Notion, PagerDuty, Sentry, Slack, and Stripe. A remote MCP server receives whatever the agent chooses to send it. None are enabled by default.
  • Nemesis8 sandboxes (optional). Running a tab in a sandbox uses a local Docker container by default. If you point Klaussy at a remote gateway instead, that tab's work runs on the host you configured.
  • Ollama (optional). Inline completion and local models run through Ollama on your own machine. No data leaves it.

2. What Klaussy stores locally

Paths below are macOS; Windows and Linux use the equivalent per-user application data directories.

  • Your project list, sessions, preferences, and MCP configuration, under ~/Library/Application Support/Klaussy/.
  • Git worktrees for each session, under ~/klaussy/sessions/. These are full checkouts of the repositories you add to a session.
  • Session notes, under ~/.klaussy/sessions/. Agents write these to hand context to each other, so they describe your architecture, schemas, and in-progress work. They sit outside your repositories and are never committed.
  • Logs of main-process activity (errors and diagnostics), in ~/Library/Logs/Klaussy/main.log. No prompt contents are logged.
  • Credentials you enter: git forge accounts, Slack and Discord webhook URLs and bot tokens, MCP server credentials, and any sandbox gateway token. These are written to Klaussy's own configuration files on disk. Klaussy does not put them in your operating system's keychain, so treat that directory as sensitive and do not sync it anywhere you would not sync a credentials file.

None of this leaves your machine unless you sync your home directory to iCloud, Dropbox, or similar.

3. What Klaussy sends off-device on its own

Two things, and both are about Klaussy itself rather than an agent acting for you:

  • Update check. Klaussy checks the GitHub releases feed for its own repository to see whether a newer version exists. That request reaches GitHub, not us. It reveals your IP address and the version you are running, and carries no personal data, no identifier we assign you, and nothing about your projects.
  • Session mirroring, when you enable it. With mirroring on, Klaussy posts that session's agent output to the Discord or Slack destination you configured, and polls it for your replies. This runs only for sessions you switch it on for.

There is no license check, no activation, no account, and no server of ours that receives your data.

Everything else that leaves your machine is a request you triggered: the agent you selected, a forge CLI Klaussy runs for you, or an MCP server you enabled. Those go to that provider, and are covered in section 1.

4. Telemetry

Klaussy does not send usage analytics or telemetry. We do not track which features you use, what projects you open, or how often you launch the app.

5. Third-party services

Klaussy orchestrates tools you choose to use. Which of these ever sees your data depends on which agents, forges, integrations, and MCP servers you turn on. Their policies apply:

Coding agents

Git forges

Session mirroring

MCP servers in the built-in catalog

None are enabled by default. A remote server only receives data once you turn it on and an agent calls it.

  • Asana, Atlassian (Jira and Confluence), Datadog, Figma, GitHub, Grafana, Honeycomb, Linear, Notion, PagerDuty, Sentry, Slack, and Stripe, each under its own provider's policy.
  • The Filesystem server is local and sends nothing off-device.

Sandboxing

  • Nemesis8, local Docker by default, or a remote gateway if you configure one.

6. Your rights

You can at any time:

  • Delete your local data by removing ~/Library/Application Support/Klaussy, ~/.klaussy, and ~/Library/Logs/Klaussy. Session worktrees under ~/klaussy/sessions are ordinary git checkouts; delete them the way you would any other clone.
  • Revoke any credential you gave Klaussy at the provider that issued it — a forge token, a Slack or Discord app, or an MCP server's credentials — and remove it from Klaussy's configuration.
  • Turn off session mirroring for a session to stop output leaving your machine for Discord or Slack.

7. Contact

Open an issue in the feedback repo for any privacy-related question.

8. Changes to this policy

We'll update this page and bump the "Last updated" date at the top. Material changes will be noted in the release notes of the version that introduces them.