Skip to content
| Marketplace
Sign in
Visual Studio Code>Other>LoreNew to Visual Studio Code? Get it now.
Lore

Lore

Matan Lasry

|
2 installs
| (0) | Free
Automated project knowledge that writes itself. Scans AI conversations and git history to generate living documentation.
Installation
Launch VS Code Quick Open (Ctrl+P), paste the following command, and press enter.
Copied to clipboard
More Info

Lore Cursor Extension

Automated project knowledge that writes itself. Scans AI conversations and git history to generate living documentation -- directly from your editor.

Prerequisites

AI Agent CLI (required -- pick one)

Lore delegates all intelligence to an AI agent CLI. Without one installed and authenticated, the extension cannot run scans.

Agent Command Install
Claude Code CLI claude docs.anthropic.com/en/docs/claude-code
Cursor Agent CLI cursor Built into Cursor IDE

Configure which agent via the lore.cliAgent setting (auto tries Claude first, then Cursor).

Other Requirements

  • Cursor IDE (VS Code ^1.105.0)
  • Node.js 18+ -- runtime (nodejs.org)
  • Git -- for scanning commit history
  • GitHub CLI (gh) -- for PR metadata and lore pr (cli.github.com)
  • Lore CLI (@lore-ai/cli) -- npm install -g @lore-ai/cli

Features

  • Automatic scans on pre-push -- triggers a background scan when you push, keeping docs up to date
  • Sidebar tree view -- browse scan status, skills, and PR descriptions from the activity bar
  • PR description generation -- generate and auto-inject PR descriptions into GitHub PRs
  • Auto-update PR on push -- optionally updates your open PR description every time you push
  • Pipeline phase runner -- run individual pipeline phases for targeted testing (dev mode)
  • Orphan detection -- find and clean up orphaned references and files
  • Agent pointer injection -- idempotent updates to AGENTS.md and CLAUDE.md
  • Conversation viewer -- browse AI conversation summaries in a webview panel
  • Worktree browser -- explore scan worktree files and open in a new window

How It Works

  1. Initialize -- run Lore: Initialize to set up ~/.lore/ and detect available AI agents
  2. Pre-push hook -- when you push, Lore automatically triggers a background scan (configurable via lore.scanOnPrePush)
  3. Background scan -- the scan runs in a terminal, summarizing conversations, classifying repos, and generating SKILL.md files
  4. Notification -- when a skill file updates, you get a notification (configurable via lore.notifyOnSkillUpdate)
  5. PR descriptions -- generate descriptions from git history and conversation summaries, inject them directly into GitHub PRs

Commands

Command Description
Lore: Initialize Set up Lore globally (~/.lore/), detect AI agents and conversation sources
Lore: Run Scan Trigger a background scan manually
Lore: Stop Scan Stop a running scan
Lore: Show Scan Terminal Open the terminal showing scan output
Lore: Open Skill Open a generated SKILL.md file
Lore: Configure Open Lore configuration settings
Lore: View Plan Details View pricing and plan information
Lore: Remove Pre-Push Hook Remove the Lore pre-push git hook
Lore: Browse Worktree Files Browse files in the scan worktree
Lore: Open Worktree in New Window Open the scan worktree in a new Cursor window
Lore: Run Phase Run an individual pipeline phase (dev mode)
Lore: Inject Agent Pointers Insert skill references into AGENTS.md and CLAUDE.md
Lore: Review Orphaned References Review and clean up dead links in generated docs
Lore: Detect Orphaned Files Find orphaned files no longer referenced by any skill
Lore: View Conversation Summary Open a conversation summary in a webview panel
Lore: Generate PR Description Generate a PR description from git history and conversation summaries
Lore: Show PR Summary Show the generated PR description summary

Configuration

Setting Default Description
lore.scanOnPrePush true Automatically trigger a scan when a pre-push hook fires
lore.notifyOnSkillUpdate true Show a notification when a skill file is updated
lore.devMode false Enable dev mode: shows individual pipeline phases in the sidebar for targeted testing
lore.cliAgent "auto" Which AI CLI agent to use for scans (see values below)
lore.costProfile "balanced" Cost profile for Lore operations. Controls which AI models are used (see values below)
lore.prDescription.enabled true Enable or disable the PR description feature entirely
lore.prDescriptionOnPush true Auto-update open PR description when pushing
lore.email "" Your email address for analytics identity and support communications
lore.telemetry true Send anonymous usage telemetry to help improve Lore

lore.cliAgent values

Value Description
auto Auto-detect: try Claude Code first, then Cursor Agent
claude Use only Claude Code CLI
cursor Use only Cursor Agent CLI

lore.costProfile values

Value Description
eco Haiku only, fastest scans, ideal for Pro tier
balanced Haiku + Sonnet, recommended for most users
quality Full model range including Opus, best for Teams/Ultra tier

Development

Building the extension

From the extension directory:

cd packages/extension
npm install
npm run build

From the project root:

npm run build:ext

Testing locally

  1. Open the packages/extension folder in Cursor
  2. Press F5 to launch the Extension Development Host
  3. The extension activates automatically in the new window

Watch mode

cd packages/extension
npm run watch

Scripts

Script Description
build Bundle the extension with esbuild
watch Bundle in watch mode for development
package Build and create a .vsix package
install-ext Install the packaged .vsix into Cursor
deploy Package and install in one step
lint Type-check with tsc --noEmit

License

MIT

  • Contact us
  • Jobs
  • Privacy
  • Manage cookies
  • Terms of use
  • Trademarks
© 2026 Microsoft