$ ssh clawdbot.space --loading...
$ ssh clawdbot.space --loading...
We reviewed 700+ ClawHub plugins and picked the 10 that are genuinely production-ready — actively maintained, security-audited, and used by thousands.
ClawHub has over 700 plugins. We installed and tested each of the top 200 by install count, ran a security audit using npx carapace scan, and monitored maintainer response times over a 90-day window. After that process, exactly 10 plugins cleared every bar we set. About 50 are reasonably good for personal use. Exactly 10 are production-grade. What separates production-grade from merely functional: the plugin has been reviewed for CVEs and supply chain attack vectors, has 1,000+ active installs with sustained positive ratings, has a maintainer who responds to issues within 72 hours, and has shipped at least one update in the last 30 days. All 10 plugins below meet every criterion. The sections after the list explain exactly how we scored them so you can evaluate future plugins using the same framework.
SQLite + FTS5 full-text search memory backend that replaces the default flat-file memory system with structured, queryable storage. Full-text search with BM25 ranking means your agent finds relevant memories even from inexact queries. Query response stays under 100ms at 100,000+ entries, making it suitable for users who've run OpenClaw for years. The automatic migration from memory.md makes initial setup nearly painless.
Headless Chrome browsing with a real browser fingerprint that passes bot detection on most commercial sites where Puppeteer-based scrapers fail. The real Chrome profile means sites that serve different content to bots get served the same experience a regular visitor would see. Essential for price monitoring, news aggregation, research, and data extraction workflows that need reliable access to JavaScript-rendered content.
Bidirectional sync with Google Calendar and Apple Calendar — creates, modifies, and cancels events using natural language commands. Shared calendar support makes it practical for household coordination: tell your agent to schedule something and it checks both your calendar and your partner's for conflicts before booking. Conflict detection prevents double-booking even across different calendar systems.
Full Notion API integration that lets your agent read, write, search, and organize your entire Notion workspace through natural language commands. Database querying with filter support means your agent can pull specific records rather than searching free-form text. Particularly popular among knowledge workers, researchers, and writers who use Notion as their primary knowledge base and want their agent to actively contribute to it.
Monitor prices across 30+ sites including Amazon, eBay, Airbnb, and hotels, with support for complex criteria that go beyond simple price thresholds. The visual AI analysis feature can assess listing photos — useful for evaluating condition, room layout, and overall appeal on marketplace listings before alerting you. Price history charts let your agent give informed context like 'this is 20% below the 90-day average.'
Drafts emails in your established writing style using tone profiles you define — formal for work contacts, casual for friends, technical for engineering colleagues. Critically, it operates in draft-only mode and never sends without explicit approval from you; this is a hard-coded safety constraint, not a configuration option. Context-aware reply suggestions analyze the incoming email thread before generating a response.
GitHub operations including automated PR review with inline comments, issue triage and labeling, code search with explanation, and release notes generation from commit history. Used by 200+ engineering teams as part of their CI/CD workflow — typically triggered when a PR is opened or when a release branch is cut. The inline comment format matches GitHub's native review system, so AI feedback appears alongside human reviewer comments.
Natural language control for your Home Assistant instance — lights, thermostats, door locks, and cameras across 200+ supported device types. Scene and automation triggers let you set up complex home behaviors through conversation rather than the HA UI. Particularly effective with TapTap AI voice integration for users who want to control their smart home hands-free while cooking, working out, or doing other activities.
Tracks expenses, monitors investment portfolios, and generates spending summaries using Plaid for bank and brokerage data. The default configuration is strictly read-only — no transactions can be initiated through the plugin — which makes it safe to connect to live accounts. Category-based spending analysis and monthly budget comparisons give your agent the context to answer financial questions intelligently rather than pointing you to an app.
Whisper-based speech-to-text and ElevenLabs text-to-speech that enables phone calls and voice-first interactions with your agent. Whisper can run locally for privacy-sensitive transcription or use the API for faster processing. ElevenLabs voice cloning lets you assign a consistent voice persona to your agent. Twilio integration handles inbound and outbound phone calls, making it possible to call your agent or have it call you with time-sensitive alerts.
| Criterion | Weight | Description |
|---|---|---|
| Security Audit | 30% | Has the plugin been reviewed for CVEs, supply chain attack vectors, and data leakage patterns? We ran npx carapace scan on every candidate and cross-referenced against the ClawHavoc malicious plugin database. |
| Active Maintenance | 25% | Last commit within 30 days, issues responded to within 72 hours on average, and at least 4 releases in the past 12 months. Abandoned plugins are a security risk — unmaintained dependencies accumulate CVEs. |
| Install Base | 20% | At least 1,000 active installs (not cumulative — active users measured by the ClawHub telemetry opt-in pool) with a sustained rating of 4.0 or higher over the trailing 90 days. |
| Documentation | 15% | Clear setup instructions with working examples, configuration option reference, and a troubleshooting section that addresses the top 5 most-reported issues. We attempted cold installation using only the docs. |
| Performance | 10% | Response time under load, memory footprint during idle operation, and token overhead per call. Plugins that add more than 500 tokens of overhead per invocation were penalized. |