Repo Intel turns your team's commits, pull requests, and file activity into answers and summaries you can read in seconds — right inside VS Code, Cursor, and other editors.

✨ Why Repo Intel?

• 💬 Ask anything, in plain English — "What did we ship this week?", "Who's touched the auth module?" — answered from real commits, PRs, and file hotspots.
• 📝 Summaries from real diffs — developer and team summaries written from the actual code changes, not just commit messages.
• 🔎 Semantic commit search — find work by intent ("when did we add rate limiting?"), powered by on‑device embeddings.
• 📅 Scheduled digests — automated standups delivered to Slack, Microsoft Teams, or Google Chat on your schedule.
• 🤖 Use the model you want — GitHub Copilot out of the box (no extra key), or bring your own Anthropic / OpenAI key.
• 🐙 Zero GitHub setup — one‑click sign‑in with VS Code's built‑in GitHub auth. No tokens to create or paste.
• 🔐 Local‑first & private — your history is cached in a private on‑device database, and semantic search runs on a bundled on‑device model (no key, no network).
• ⚡ Three ways to ask — the sidebar chat, the Command Palette, or @repoIntel in Copilot Chat.

🚀 Get started

1. Install and open the Repo Intel tab in the activity bar.
2. Sign in to GitHub — one click, no tokens.
3. Sync, then ask away: "Summarize my work this month."

The workspace repo is tracked automatically — use Choose repos in the panel header to add any others you have access to.

🧠 Models & privacy

Two things are always on your machine — no API key, no network:

• the commit / PR cache (a local SQLite database in the extension's storage), and
• the semantic‑search index (embeddings computed by a small bundled model).

The AI answers and summaries, however, need a language model. Pick one from the Provider dropdown in the panel header:

⚙️ Settings

Model provider, tone, and auto‑sync are also switchable from buttons in the panel header.

🔒 Telemetry

Repo Intel collects anonymous usage and error telemetry to help improve the extension. It records only feature‑usage categories and error types (e.g. tool_call, answer_truncated, an error's class name) plus the extension version — never any personally identifiable information, your code, commit content, repository names, prompts, answers, or API keys.

Your choice, asked once. On first launch Repo Intel asks whether you want to share anonymous telemetry, and remembers your answer:

• Allow → telemetry is sent. By default this still respects your global telemetry.telemetryLevel (so global off means nothing is sent).
• No thanks → telemetry is never sent, even if your global telemetry is on.
• If you never answer → Repo Intel falls back to your VS Code telemetry.telemetryLevel setting (off ⇒ nothing is sent).

In short: Repo Intel never sends telemetry when your VS Code telemetry is off, and your explicit "No thanks" is always honored. Change your mind anytime with the Repo Intel: Telemetry Settings command, and see exactly what would be sent in the Repo Intel output channel (View → Output → "Repo Intel").

💬 Feedback

Run Repo Intel: Send Feedback from the Command Palette to open a pre‑filled GitHub issue (bug report, feature request, or general feedback). Nothing is sent automatically — it just opens your browser to the issue form.