From lgtm
Address review comments from the lgtm UI. Use when asked to handle code review feedback.
How this skill is triggered — by the user, by Claude, or both
Slash command
/lgtm:lgtmThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
When invoked, enter a review loop that runs until the session is approved or abandoned.
When invoked, enter a review loop that runs until the session is approved or abandoned.
Loop:
lgtm fetch — blocks until the developer clicks "Submit to agent" in the UIsession_status in the output. If approved or abandoned, report and stopopen_threads is empty, report "No open threads to address" and go to step 1file, line_start, line_end)comments to understand what the developer is askinglgtm reply <thread_id> "explanation of what you changed"lgtm thread --file <path> --line <n> --severity <level> "description"
Do not fix agent-raised issues unless the developer explicitly asks# Start a review session (launches lgtm app if not running)
lgtm start --base main
# Wait for developer to submit comments (blocks until ready)
lgtm fetch
# Reply to a thread after fixing the code
lgtm reply t_01J8XYZABC "Added exponential backoff with jitter. See lines 42-67."
# Raise an observation for the developer to review
lgtm thread --file src/auth.py --line 71 --severity warning "Hardcoded API key — should use environment variable"
# Check session status without blocking
lgtm status --json
# Approve the session (all threads must be resolved)
lgtm approve
# Abandon the session
lgtm abandon
# Remove a session
lgtm clean
The lgtm app is a persistent Tauri desktop application with multiple tabs (one per review session). The CLI is a thin HTTP client that communicates with the app via its local API. No per-session servers — one app handles all reviews.
lgtm start will launch it automaticallylgtm start firstlgtm reply with an invalid thread ID: skip and continue with remaining threadsProvides behavioral guidelines to reduce common LLM coding mistakes, focusing on simplicity, surgical changes, assumption surfacing, and verifiable success criteria.
Searches, retrieves, and installs Agent Skills from prompts.chat registry using MCP tools like search_skills and get_skill. Activates for finding skills, browsing catalogs, or extending Claude.
Creates, edits, and optimizes skills for Claude Code, including drafting, evaluating with test prompts, iterating on performance, and improving skill descriptions for better triggering accuracy.
npx claudepluginhub yumike/lgtm --plugin lgtm