lrc - LiveReview CLI

May 26, 2026 · View on GitHub

lrc is a command-line tool for submitting local code diffs to LiveReview for AI-powered code review.

What data is sent?

Only the staged diff bundle required for review is sent to LiveReview APIs. No full repository context is uploaded.

git-lrc now uses explicit boundaries for data handling:

storage/*: local file/DB persistence only (config, hook metadata, attestation DB/files, review outputs, self-update state files)
network/*: outbound HTTP calls only (review submit/poll, connector APIs, auth refresh, provisioning, self-update manifest/binary fetch)

Transmission behavior:

Review submission sends a zip/base64 of the selected diff (/api/v1/diff-review)
Optional telemetry sends CLI usage ping (/api/v1/diff-review/cli-used)
Connector setup/validation calls go to connector/auth endpoints when you run setup/UI flows

Local persistence behavior:

Credentials and connector state are written to ~/.lrc.toml
Hook/attestation artifacts are written under the repository git dir at $(git rev-parse --git-dir)/lrc/...
Optional review artifacts are written only when you pass --save-* flags

Retention and deletion:

Local files are user-controlled and can be deleted directly from ~/.lrc.toml, ~/.lrc/update/*, and $(git rev-parse --git-dir)/lrc/*
Server-side retention/training policy is governed by LiveReview backend policy; the CLI does not independently retain uploaded diff payloads after submission

Data Handling Boundaries

To make confidentiality behavior explicit, the CLI uses two boundary packages:

storage/*: all local persistence (config files, hook/attestation files, sqlite mutations, output artifacts, self-update state)
network/*: all outbound HTTP traffic (setup/provisioning, review submit/poll, connector proxy/auth refresh, self-update fetches)

This gives one place to audit what is written locally vs what leaves the machine.

Installation

Build the binary:

make build-local

Or build directly:

go build -o lrc .

Usage

Basic Usage

Review staged changes:

lrc --api-key YOUR_API_KEY

Git Hook Installation

Install pre-commit hooks for automatic code review:

# Install hooks
lrc hooks install

# Uninstall hooks
lrc hooks uninstall

# Force update existing lrc hooks
lrc hooks install --force

The hooks will:

Run lrc review --staged before each commit
Never block commits (always exits 0)
Can be skipped with Ctrl-C
Can be bypassed with git commit --no-verify
Add a commit trailer: LiveReview Pre-Commit Check: [ran|skipped]

Diff Sources

Staged changes (default):

lrc --api-key YOUR_API_KEY --diff-source staged

Working tree changes:

lrc --api-key YOUR_API_KEY --diff-source working

Git range:

lrc --api-key YOUR_API_KEY --diff-source range --range HEAD~1..HEAD

From file:

lrc --api-key YOUR_API_KEY --diff-source file --diff-file my-changes.diff

Configuration

You can provide configuration in three ways (in order of precedence):

Command-line flags: --api-key YOUR_API_KEY --api-url https://your-instance.com
Environment variables: export LRC_API_KEY="your-api-key" LRC_API_URL="https://your-instance.com"
Config file: Create ~/.lrc.toml with:
```
# Your LiveReview API key
```

api_key = "lr_example_token"

Your LiveReview API endpoint (base URL, without /api suffix)

The CLI automatically appends /api/v1/diff-review

api_url = "https://manual-talent.apps.hexmos.com"


#### Setting Up Your Config File

```bash
# Copy the sample config
cp cmd/lrc/.lrc.toml.sample ~/.lrc.toml

# Edit with your credentials
vim ~/.lrc.toml

# Protect your API key
chmod 600 ~/.lrc.toml

# Now you can run lrc without any flags
lrc

All other flags can be set via environment variables or command-line flags.

Flags

Flag	Environment Variable	Default	Description
`--repo-name`	`LRC_REPO_NAME`	current dir basename	Repository name
`--diff-source`	`LRC_DIFF_SOURCE`	`staged`	Diff source: `staged`, `working`, `range`, or `file`
`--range`	`LRC_RANGE`		Git range (e.g., `HEAD~1..HEAD`) for `range` mode
`--diff-file`	`LRC_DIFF_FILE`		Path to diff file for `file` mode
`--api-url`	`LRC_API_URL`	`http://localhost:8888`	LiveReview API base URL
`--api-key`	`LRC_API_KEY`	(from config)	API key for authentication
`--poll-interval`	`LRC_POLL_INTERVAL`	`2s`	Interval between status polls
`--timeout`	`LRC_TIMEOUT`	`5m`	Maximum wait time for review
`--blocking-review`	`LRC_BLOCKING_REVIEW`	`false`	Launch the decision-capable web review UI and block until a proceed or abort decision is made
`--blocking-review-timeout`	`LRC_BLOCKING_REVIEW_TIMEOUT`	`20m`	Maximum total time blocking review mode may hold the caller before aborting
`--output`	`LRC_OUTPUT`	`pretty`	Output format: `pretty` or `json`
`--save-bundle`	`LRC_SAVE_BUNDLE`		Save bundle to file for inspection before sending
`--save-json`	`LRC_SAVE_JSON`		Save JSON response to file after completion
`--save-text`	`LRC_SAVE_TEXT`		Save formatted text with comment markers to file
`--save-html`	`LRC_SAVE_HTML`		Save GitHub-style HTML review to file
`--verbose, -v`	`LRC_VERBOSE`	`false`	Enable verbose output

Examples

Setup API key in config file

# Create config file with your API key and endpoint
cat > ~/.lrc.toml << EOF
# Your LiveReview API key
api_key = "lr_example_token"

# Your LiveReview API endpoint
api_url = "https://manual-talent.apps.hexmos.com"
EOF

# Protect your config file
chmod 600 ~/.lrc.toml

# Now you can run lrc without specifying the key or URL
lrc

Inspect bundle before sending

# Save the bundle for inspection before submitting
lrc --save-bundle bundle.txt

# Review the bundle file to see what will be sent
less bundle.txt

Save review results to files

# Save both JSON and formatted text output
lrc --save-json review.json --save-text review.txt

# Save HTML output for viewing in browser
lrc --save-html review.html

# Save all formats
lrc --save-json review.json --save-text review.txt --save-html review.html

# Search for comments in the text file
grep -n ">>>COMMENT<<<" review.txt

# Or use your editor's search function to jump between comments
vim review.txt  # then search for: />>>COMMENT<<<

# Open HTML review in browser
xdg-open review.html  # Linux
open review.html      # macOS
start review.html     # Windows

Complete workflow with all output options

# Run review with all inspection/output options
lrc \
  --save-bundle bundle.txt \
  --save-json review.json \
  --save-text review.txt \
  --save-html review.html \
  --verbose

# Review the bundle that was sent
cat bundle.txt

# Check the raw JSON response
jq . review.json

# Navigate through comments in text file
# Search for ">>>COMMENT<<<" to jump between comments
vim review.txt

# Open HTML review in browser (best for visual inspection)
xdg-open review.html

Review uncommitted changes

# Review all staged changes
git add .
lrc --api-key YOUR_API_KEY

# Review working directory changes (unstaged)
lrc --api-key YOUR_API_KEY --diff-source working

Review a specific commit

lrc --api-key YOUR_API_KEY --diff-source range --range HEAD~1..HEAD

Review a saved diff

git diff main..feature-branch > changes.diff
lrc --api-key YOUR_API_KEY --diff-source file --diff-file changes.diff

JSON output for scripting

lrc --api-key YOUR_API_KEY --output json > review-results.json

Verbose mode for debugging

lrc --api-key YOUR_API_KEY --verbose

Output Formats

Pretty (default)

Human-readable output with file sections and colored severity levels:

================================================================================
LIVEREVIEW RESULTS
================================================================================

Summary:
The code looks good overall. Minor suggestions below.

2 file(s) with comments:

--------------------------------------------------------------------------------
FILE: src/main.go
--------------------------------------------------------------------------------

  [WARNING] Line 42 (best-practices)
    Consider using context.WithTimeout instead of time.Sleep

  [INFO] Line 89 (style)
    Variable name could be more descriptive

================================================================================
Review complete: 2 total comment(s)
================================================================================

JSON

Machine-readable JSON output for automation:

{
  "status": "completed",
  "summary": "The code looks good overall. Minor suggestions below.",
  "files": [
    {
      "file_path": "src/main.go",
      "hunks": [...],
      "comments": [
        {
          "line": 42,
          "content": "Consider using context.WithTimeout instead of time.Sleep",
          "severity": "warning",
          "category": "best-practices"
        }
      ]
    }
  ]
}

API Requirements

The lrc tool requires:

A running LiveReview API server (default: http://localhost:8888)
A valid API key (obtain from your LiveReview account settings)

The tool communicates with these endpoints:

POST /api/v1/diff-review - Submit diff for review
GET /api/v1/diff-review/:id - Poll for review status/results

Exit Codes

0 - Success
1 - Error (network failure, invalid input, review failed, timeout, etc.)

Troubleshooting

"API key not provided"

You can provide the API key in three ways:

# Option 1: Config file (recommended)
echo 'api_key = "your-key"' > ~/.lrc.toml

# Option 2: Environment variable
export LRC_API_KEY="your-key"

# Option 3: Command-line flag
lrc --api-key your-key

Inspecting what gets sent to the API

Use --save-bundle to see exactly what will be transmitted:

lrc --save-bundle bundle.txt --verbose

# The bundle file contains:
# - Original diff content
# - Zip archive info
# - Base64 encoded payload (what the API receives)

Analyzing review comments

Use --save-text to get a searchable text file with markers:

lrc --save-text review.txt

# Search for comments:
grep ">>>COMMENT<<<" review.txt

# Or use editor search to jump between comments:
vim review.txt  # then: />>>COMMENT<<<
code review.txt  # then: Ctrl+F ">>>COMMENT<<<"

"no diff content collected"

Make sure you have uncommitted changes:

git status
git diff --staged  # Check if there are staged changes

"API returned status 401"

Check your API key:

echo $LRC_API_KEY
# or
lrc --api-key YOUR_API_KEY --verbose

"timeout waiting for review completion"

Increase the timeout:

lrc --api-key YOUR_API_KEY --timeout 10m

Connection refused

Ensure the LiveReview API server is running:

# Start the API server
./livereview api

# Or check if it's already running
curl http://localhost:8888/health