Claude Code Plugins by alanceloth

A collection of plugins for Claude Code — Anthropic's official CLI for Claude.

Available

Statusline

2-line status bar showing context usage, cost, session time, git info, system stats, and rate limit usage.

[Opus] my-repo | main +2 ~1 | RAM:8.2G/16G(51%) CPU:12%
[====------] 42%  $0.05  8m5s  |  5h:23% 7d:41%

Quick install:

git clone https://github.com/alanceloth/claude-code-plugins.git && bash claude-code-plugins/statusline/install.sh

See statusline/README.md for details.

---

Plugins

PR Review Orchestrator

Meta-orchestrator that dynamically discovers all installed review specialist agents, dispatches them in parallel, scores findings for confidence, and aggregates the survivors into a single consolidated PR comment.

Features:

• Dynamic agent discovery — automatically detects all installed review agents (ring-default, pr-review-toolkit, comprehensive-review, coderabbit, etc.)
• Context-aware selection — analyzes diff patterns, file types, PR title/body to pick the right agents
• Parallel dispatch — launches all selected agents simultaneously for fast reviews
• Eligibility gate (v1.1) — skips closed/draft/automated PRs and ones we already reviewed recently
• CLAUDE.md awareness (v1.1) — loads repo conventions and injects them into reviewers so they enforce your standards
• Confidence scoring (v1.1) — each finding is scored 0-100 by a Haiku pass; cuts noise below --min-confidence (default 75)
• Modified-line filter (v1.1) — drops findings on lines the PR didn't touch (kills pre-existing-code noise)
• Comment dedup (v1.1) — suppresses findings already raised in earlier reviews of the same PR
• Permalink citations (v1.1) — every finding links to the exact line at the head SHA, surviving rebase/squash
• Smart aggregation — deduplicates findings, sorts by severity, merges overlapping issues
• Bilingual output — auto-detects Portuguese or English from PR body
• Size-aware scaling — adjusts agent count based on PR size
• Token-disciplined — Haiku handles all mechanical phases (eligibility, scoring, dedup, formatting); Sonnet+/Ring agents reserved for semantic review

Install

# 1. Add the marketplace
/plugin marketplace add alanceloth/claude-code-plugins

# 2. Install the plugin
/plugin install pr-review-orchestrator@alanceloth-plugins

Usage

# Review a specific PR
/pr-review-orchestrator:pr-review 42

# Auto-detect PR from current branch
/pr-review-orchestrator:pr-review

# Preview which agents would be dispatched
/pr-review-orchestrator:pr-review 42 --dry-run

# Force Portuguese output
/pr-review-orchestrator:pr-review 42 --lang pt

# Review without posting to GitHub
/pr-review-orchestrator:pr-review 42 --no-post

# Skip specific agents
/pr-review-orchestrator:pr-review 42 --skip code-simplifier,comment-analyzer

# Stricter review (only findings scoring >=85/100)
/pr-review-orchestrator:pr-review 42 --min-confidence 85

# Looser review (catch more, accept more noise)
/pr-review-orchestrator:pr-review 42 --min-confidence 60

# Re-review a PR you already reviewed
/pr-review-orchestrator:pr-review 42 --skip-eligibility

# Skip CLAUDE.md loading (raw review against generic best practices)
/pr-review-orchestrator:pr-review 42 --skip-claude-md

# Include findings on pre-existing code (lines not touched by the diff)
/pr-review-orchestrator:pr-review 42 --include-preexisting

The skill also auto-triggers when you say things like "review my PR" or "revisar o PR".

License

MIT