Skip to content

PRIVACY.md

Latest commit

 

History

History
73 lines (51 loc) · 3.65 KB

PRIVACY.md

File metadata and controls

73 lines (51 loc) · 3.65 KB
Raw

Privacy Policy — Live Memory (Claude Code plugin)

Last updated: 2026-07-05

Live Memory is an open-source Claude Code plugin published by shofer.dev. It runs entirely on your own machine as a local MCP server. This policy describes exactly what data it touches, what leaves your machine, and what is stored — so you can decide with full information.

Short version: the plugin author (shofer.dev) collects nothing. There is no telemetry, no analytics, and no server operated by shofer.dev that receives your data. The only data that leaves your machine goes to the LLM provider you configure (Anthropic by default), under that provider's terms.

What the plugin accesses

To answer your agent's codebase questions, Live Memory reads source files in the workspace (repo) you point it at:

  • Passively: hooks tee the content of files your Claude Code agent already reads or edits to the local server, so it learns from work you were doing anyway (no extra reading).
  • Actively: when answering a question it can read files itself using read-only, path-jailed tools (Grep/Read) confined to the workspace. It cannot edit, create, delete, or execute anything.

It does not access files outside the workspace, your environment secrets, your browser, or any other application.

What leaves your machine, and to whom

To generate an answer, Live Memory sends the relevant codebase content and your question to the LLM provider you have configured — and to no one else:

  • Default: Anthropic (api.anthropic.com), using your existing Claude subscription via an OAuth token (it also contacts Anthropic's OAuth endpoint, console.anthropic.com, to refresh that token). No API key is required in this mode.
  • If you reconfigure it (via /live-memory-config or environment variables): the provider you choose instead — e.g. DeepSeek or any OpenAI-compatible endpoint, or a fully local model (in which case nothing leaves your machine at all).

That data is processed by the chosen provider under their privacy policy and your account terms with them (for Anthropic, see Anthropic's Privacy Policy and Commercial/Consumer Terms). shofer.dev is not a party to that transfer and never receives a copy.

No third-party analytics or telemetry are included. The plugin makes no network calls other than to the configured LLM provider (and Anthropic's OAuth endpoint for subscription auth).

What is stored, and where

Live Memory keeps its accumulated codebase knowledge as JSON snapshot files on your local disk (by default under ~/.claude/plugins/data/live-memory/, or wherever you set LIVE_MEMORY_DATA_DIR / CLAUDE_PLUGIN_DATA). This is local persistence so the memory survives across sessions. It is never uploaded anywhere. Raw teed file bytes are held in memory only; snapshots keep a distilled summary and file manifest, not full file contents.

Your controls

  • Wipe the memory for a workspace (or all) at any time: /live-memory-empty.
  • Inspect what it holds and its cost: /live-memory-stats.
  • Change or localize the model/provider (including running fully offline): /live-memory-config.
  • Stop it by stopping the local server; remove all stored data by deleting the data directory above; uninstall the plugin with /plugin uninstall.

Children

Live Memory is a developer tool and is not directed to children under 13.

Changes

Material changes to this policy will be reflected in this file in the plugin's repository, with an updated date.

Contact

Questions or concerns: open an issue at https://github.com/shofer-dev/claude-code-live-memory/issues or contact shofer.dev.