Lore — AI Codebase Intelligence
Your codebase has a story. Now you can read it.
Ask plain English questions about any codebase and get accurate, sourced answers — without sending a single line of code to the cloud. Lore runs entirely on your own hardware using local AI models via Ollama.
Privacy-first. Air-gap compatible. No API keys. No telemetry. Ever.
Why Not Just Use ChatGPT?
|
ChatGPT / Claude |
Lore |
| Privacy |
Your code goes to the cloud |
Runs 100% locally — nothing leaves your machine |
| Scale |
Limited context window |
Handles thousands of files via semantic search |
| Output |
Conversation |
Structured: diagrams, CVE reports, impact analysis |
| Stays current |
Manual copy-paste every time |
Git hooks + webhooks re-index on every commit |
| Setup |
Paste and hope |
Point at a repo once, ask questions forever |
Features
Ask Questions About Your Codebase
Type any question in the Lore sidebar and get a streaming, word-by-word answer sourced from your actual code — with file references.
- "How does authentication work?"
- "Where is the payment processing logic?"
- "What calls this function?"
In-Editor Analysis
- Smell score — risk score for the file you're currently editing
- Change impact — what breaks if you change the current file
- Architecture diagrams — interactive module, dependency, and data flow diagrams inside VS Code
Full Analysis Suite (via CLI)
- Dead code detection — functions defined but never called
- Dependency audit — CVE checks on every package, CycloneDX SBOM generation
- Change impact analysis — downstream blast radius before you touch a line
- Deep file explanation — call graph, smell score, callers, suggestions
- Onboarding guide — tailored doc for new engineers joining a codebase
- Audience briefs — different summaries for engineers, execs, auditors, acquirers
- Architecture diagrams — module overview, call flow, data flow (Mermaid/SVG/PNG)
- PR analysis — impact, smells, and dependency changes for any pull request
- Git changelog — plain English changelog between any two refs
- Decision log — log and query architectural decisions (ADRs)
- Health tracking — track smells, CVEs, and quality scores over time
Keyboard Shortcuts
| Shortcut |
Action |
Ctrl+Shift+L |
Ask a question about the codebase |
Ctrl+Shift+Alt+L |
Ask about selected code |
Select any code and right-click to ask Lore about it directly from the editor context menu.
Setup
Step 1 — Install Ollama
Download from ollama.com. Lore will automatically detect your hardware and pull the right model on first run.
Step 2 — Install the Lore CLI
pip install lore-cli
Step 3 — Try the demo (optional but recommended)
lore demo
Runs a guided walkthrough against a bundled demo codebase — no setup required. See exactly what Lore can do in about 3 minutes.
Step 4 — Index your codebase
lore init --path /path/to/your/codebase
Auto-detects your stack, indexes your codebase, and installs a git hook to keep the index current on every commit.
First time? Run lore doctor to check all dependencies are in place.
Step 5 — Open VS Code
- Install this extension
- Open your codebase as a workspace folder
- Click the Lore icon in the activity bar
- Start asking questions
Requirements
- Ollama — runs the local AI model (auto-detected and configured on first run)
- Python 3.12+ with
pip install lore-cli
- Or: Docker with
docker compose up lore-server -d
Configuration
| Setting |
Default |
Description |
lore.serverUrl |
http://localhost:8000 |
URL of the Lore server |
lore.cliPath |
lore |
Path to the lore CLI executable |
lore.codebasePath |
(workspace root) |
Override the codebase path |
lore.preferServer |
true |
Use server if available, fall back to CLI |
lore.autoSmellOnSave |
false |
Show smell score on every file save |
Privacy
Lore runs entirely on your own hardware. No code ever leaves your network.
- No cloud API calls — the AI model runs via Ollama locally
- No telemetry or usage data — Lore does not phone home
- Air-gap compatible — works with zero internet connectivity
- On-premise vector database — embeddings stored locally in ChromaDB
- CVE checks use a local database — no outbound security API calls
Suitable for defense contractors, healthcare, government, and any organization with strict data compliance requirements.
Full CLI Reference
lore ask "how does auth work?" # Plain English Q&A
lore find "stock reorder logic" # Semantic code search
lore smell --path . # Code quality scoring
lore dead --path . # Dead code detection
lore deps --format html # Dependency audit + CVE report
lore diagram --type module_overview # Architecture diagram
lore impact --file core/auth.py # Change impact analysis
lore explain --file core/auth.py # Deep file explanation
lore onboard --role backend # Onboarding guide
lore brief --for exec # Audience-tailored summary
lore map --path . # Full dependency graph
lore changelog --last 10 # Plain English git changelog
lore pr --platform github --repo o/r --pr 42 # PR analysis
lore decide add --title "..." --reason "..." # Log an ADR
lore remember --save # Save health snapshot
lore demo # Guided feature walkthrough
License
Business Source License 1.1 — free for individual, non-commercial use.
Converts to Apache 2.0 on April 2, 2030.
Team and enterprise licensing: get-lore.com
Built by SmithBuilds LLC