LGTM CLI

AI code review + CI/CD security, from your terminal.

The official CLI for LGTM (Looks Good To Meow) — review code, manage repos, trigger security scans, mint API tokens, and inspect analytics from the command line. JSON output everywhere, real-time agent progress, and structured error codes for scripting.

$ npm install -g @tarin/lgtm-cli
$ lgtm login

What you get

• lgtm review — review your local diff or a GitHub PR with the same 6-agent pipeline + synthesizer as the dashboard. Real-time progress via SSE.
• lgtm pr — list, view, trigger, or open reviews on existing GitHub PRs.
• lgtm security — full LGTM Security surface: enroll repos, view audit log, set per-rule policy, mint API tokens for the runtime watchdog GitHub Action.
• lgtm repo — connect, index, configure, list, and inspect repos. Per-repo auto-review, focus areas, and provider overrides.
• lgtm config — manage AI provider keys (OpenAI / Anthropic / Gemini) and default model.
• lgtm doctor — diagnose setup problems with a single command.
• lgtm notifications / lgtm analytics — read-only insight into reviews and findings.
• --json on every read command — pipe into jq, scripts, or CI.
• Shell completion — lgtm completion bash|zsh|fish.
• Auto-update notice — passive once-a-day check.

Quickstart

# 1. Install
npm install -g @tarin/lgtm-cli

# 2. Log in (opens your browser)
lgtm login

# 3. Verify everything is wired up
lgtm doctor

# 4. Add an AI provider key
lgtm config set-key --provider openai --key sk-...

# 5. Connect your current repo + build the context index
cd /path/to/your/repo
lgtm repo connect
lgtm repo index

# 6. Run your first review
lgtm review

Commands

Auth

| Command | Description | |---|---| | lgtm login [--api-url <url>] | OAuth via GitHub. Pass --api-url for staging. | | lgtm logout [-y] | Clear the local session. | | lgtm whoami [--json] | Show user, plan, usage, AI config, token expiry. | | lgtm doctor [--json] | Run all setup checks and report what's broken. |

Code review

| Command | Description | |---|---| | lgtm review | Review local uncommitted diff against the detected default branch. | | lgtm review --staged | Review only the staged diff. | | lgtm review --base develop | Compare against an arbitrary base ref. | | lgtm review --pr 42 | Trigger + stream a review on PR #42. | | lgtm review --json | Emit machine-readable JSON. | | lgtm pr list [--mine] [--repo X] [--status reviewing] | List PRs reviewed by LGTM. | | lgtm pr view <num> | Show a PR's review (verdict, summary, agent breakdown). | | lgtm pr review <num> | Alias for lgtm review --pr. | | lgtm pr open <num> | Open the PR's report in your browser. |

LGTM Security

| Command | Description | |---|---| | lgtm security enroll [--repo X] | Enroll a repo in LGTM Security. | | lgtm security list | List enrolled repos with open-finding counts. | | lgtm security status [--repo X] | Show a repo's enrollment + posture. | | lgtm security scan [--repo X] | Trigger a manual scan. | | lgtm security audit [--open] [--severity high] [--rule-id ...] [--pr 42] [--query "..."] | Filter the audit log. | | lgtm security resolve <id> --resolution fixed\|muted\|false-positive | Resolve an audit entry. | | lgtm security policy show | Display per-rule policy. | | lgtm security policy set <rule> <block\|warn\|off> | Override a rule's action. | | lgtm security tokens list | List API tokens. | | lgtm security tokens create [name] [--expires-in-days N] | Mint a token for the watchdog Action. | | lgtm security tokens revoke <id> | Revoke a token. | | lgtm security open | Open the LGTM Security dashboard page. |

Repo

| Command | Description | |---|---| | lgtm repo connect [--repo X] | Install GitHub App + connect a repo. | | lgtm repo status [--repo X] | Show installation, connection, index, and config state. | | lgtm repo list | Table of all connected repos with index status + provider override. | | lgtm repo disconnect [-y] [--repo X] | Remove from LGTM. | | lgtm repo index [--repo X] | Rebuild the tree-sitter context index. | | lgtm repo settings [--auto-review true] [--focus-areas bugs,security] [--provider openai] [--model gpt-5.4] [--pr-chat true] | View or update per-repo settings. | | lgtm repo sync [--repo X] | Re-sync PRs from GitHub. | | lgtm repo open [--repo X] | Open the repo's dashboard page. |

Config

| Command | Description | |---|---| | lgtm config set-key --provider openai\|gemini\|anthropic --key <KEY> | Save and validate a provider key. | | lgtm config set-model --provider X [--model Y] | Set default model (interactive if model omitted). | | lgtm config show | Show current configuration with masked keys. | | lgtm config remove-key --provider X | Remove a saved key. |

Analytics & notifications

| Command | Description | |---|---| | lgtm analytics overview [--range 7d\|30d\|90d\|all] [--repo X] | Totals, verdict distribution, severity breakdown. | | lgtm analytics top-issues [--range R] [--repo X] | Most common findings. | | lgtm analytics trends [--range R] [--repo X] | Time series (PRs/week, review time, severity). | | lgtm notifications list [--unread] [--limit N] | Read in-app notifications. | | lgtm notifications read <id> | Mark one as read. | | lgtm notifications read-all | Mark every notification as read. |

Open

| Target | Opens | |---|---| | lgtm open (default dashboard) | /dashboard | | lgtm open repos | /dashboard/repos | | lgtm open security | /dashboard/security | | lgtm open settings | /dashboard/settings | | lgtm open analytics | /dashboard/analytics | | lgtm open pricing | /dashboard/pricing | | lgtm open docs | Public docs site |

Shell completion

# Bash
lgtm completion bash > ~/.lgtm.bash && echo "source ~/.lgtm.bash" >> ~/.bashrc

# Zsh
lgtm completion zsh > ~/.lgtm.zsh && echo "source ~/.lgtm.zsh" >> ~/.zshrc

# Fish
lgtm completion fish > ~/.config/fish/completions/lgtm.fish

JSON output

Every read command (and most writes) accepts --json. Errors emit structured JSON to stderr with a stable code field. Example:

$ lgtm whoami --json | jq '.billing.reviewsRemaining'
50

$ lgtm pr list --json --status reviewing | jq '.prs[] | .prNumber'

$ lgtm doctor --json | jq '.summary'
{
  "total": 9,
  "pass": 9,
  "warn": 0,
  "fail": 0,
  "skip": 0,
  "healthy": true
}

Error codes (stable across versions): not_logged_in, no_repo, not_connected, not_indexed, not_configured, billing_limit, rate_limited, invalid_input, server_error, network_error, auth_expired, not_found, conflict, permission_denied, unknown.

Configuration

The CLI stores its config in ~/.lgtm/:

• config.json — login token, API URL, expiry. Created by lgtm login.
• .env (optional) — environment-variable overrides loaded at every command. Useful for LGTM_API_URL.

You can also pass --api-url to lgtm login to target a staging server.

Development

# Build
npm install
npm run build

# Run unit + integration tests
npm test

# Run the built CLI directly
node bin/lgtm.js --help

Requirements

• Node 18+
• git
• A GitHub account with the LGTM GitHub App installed on at least one repo
• An API key from OpenAI (today) or Gemini / Anthropic (coming soon)

Links

• Website: https://looksgoodtomeow.in
• Docs: https://looksgoodtomeow.in/docs
• GitHub App: https://github.com/apps/tarin-lgtm/installations/new
• Issues / support: [email protected]