Browse plugins

Hookify creates custom hooks to prevent unwanted behaviors by analyzing conversation patterns or explicit instructions, using simple markdown configuration files with regex pattern matching. It supports four hook events (PreToolUse, PostToolUse, UserPromptSubmit, Stop) and provides commands to generate rules, list configurations, and interactively manage hook behavior without requiring restarts.

Pro Workflow is a complete AI coding system with self-correcting memory, persistent SQLite-indexed research wikis, and an auto-research loop that learns from corrections across sessions. It provides 41 skills, 8 agents, 23 commands, and intelligent hooks that inject relevant knowledge and enforce quality gates without requiring API keys or external services.

Claude-mem is a memory compression system that persists context across Claude Code sessions through 19 skills and an MCP server, using shell commands and hooks to manage context retention. It integrates with the Code editor's session lifecycle via pre/post tool hooks and prompt submission events.

This plugin hooks into Claude Code's session start and user prompt submission events to capture usage metrics, rejections, and session statistics. It collects telemetry data and displays it in a unified dashboard for monitoring plugin performance and user behavior patterns.

Claude-Mem for Cowork captures tool use and prompts via hooks, streams them to cmem.ai for server-side observation synthesis, and injects compiled context back into new sessions and spawned agents. It provides fail-soft operation with spool fallback, a mem-search skill for querying memory, and automatic project-based naming in ephemeral cloud sessions.

Ars Contexta generates a personalized knowledge management system through conversational setup, creating markdown vaults with wiki links, processing pipelines, automation hooks, and domain-native documentation—all derived from how you describe your thinking and work. The plugin provides 10 skills for extracting insights, finding connections, and maintaining your vault, plus a 6-phase pipeline that mirrors Cornell note-taking with a meta-cognitive layer for continuous system evolution.

Ponytail is a Claude Code plugin that enforces minimal, production-ready solutions by prioritizing YAGNI principles, standard library usage, and avoiding unnecessary abstractions. It reduces code bloat, cost, and execution time while maintaining full safety guarantees through session-start and prompt-submit hooks that guide the agent toward simpler implementations.

AutoHarness is a self-learning skill layer for Claude Code that automatically distills reusable skills from real work sessions, consolidates similar skills to avoid duplication, and prunes unused ones to keep the library clean. Skills are validated by actual adherence in use rather than offline benchmarks, and the system maintains a per-skill ledger of decisions and scenarios for transparency.

This plugin provides five core skills for ecosystem navigation, skill creation, research pattern analysis, session reflection, and plugin development, with a UserPromptSubmit hook that forces skill evaluation on user input. It integrates via a shell-based activation script to enable automated skill assessment workflows.

memsearch automatically captures and recalls Claude's work across sessions by hooking into Claude Code's lifecycle and storing summarized memories as searchable embeddings. The plugin runs entirely locally using ONNX-based embeddings, requiring no API keys or external services.

A Claude Code skill that reformats assistant output for ADHD-friendly readability by leading with actionable next steps, numbering multi-step tasks, and eliminating tangents and filler language. It applies 10 structured rules (e.g., no preamble, specific time estimates, suppressed asides) to make technical guidance direct and scannable.

Hivemind is a Claude Code plugin that provides cloud-backed persistent memory using Deeplake, allowing you to read, write, and share memory across multiple Claude Code sessions and agents through hooks and skills. It enables long-term state management and data persistence without requiring external API keys, using Node.js as the runtime dependency.

This plugin intercepts write operations and blocks a predefined list of AI buzzwords before they're used, helping developers maintain more authentic language in their code and documentation. It leverages Python-based hooks to filter and flag overused terminology in real time.

This Claude Code plugin synchronizes CLAUDE.md documentation and a permissions allowlist with an external Python script, then refreshes context on demand via a PostToolUse hook. It provides four commands to manage plugin metadata and permissions without requiring external API keys.

This skill teaches Claude Code agents when and how to use dedicated structured data tools (jq, yq, dasel, qsv, mlr) instead of fragile text processing on JSON, YAML, TOML, XML, CSV, and JSONL formats. It includes a PreToolUse hook that guides tool selection and provides comprehensive cookbooks for each tool's patterns and use cases.

Ralph Wiggum implements continuous self-referential AI loops that repeat the same prompt through a Stop hook until task completion, allowing iterative refinement within a single Claude Code session. The technique runs Claude in a while-true pattern with persistent file state, letting the model read its own past work and fix failures autonomously.

This plugin adds audio feedback to Claude Code on Windows by playing system sounds when Claude finishes responding or when notifications occur using PowerShell. It provides non-blocking background playback and is configurable via environment variables to select different Windows system sounds or disable the feature entirely.

This plugin hooks into the PreCompact event to automatically summarize file paths, root causes, and open questions, keeping context intact during code compaction. It relies on an external shell script (precompact_priorities.sh) to process and organize information before compaction occurs.

This Agent Skill detects missing CLI tools from error patterns, audits project environments, and installs them automatically via optimal package managers (binary downloads, Cargo, pip, npm, or system packages). It supports 77+ tools across development stacks including Python, Node.js, Rust, Go, Docker, and DevOps platforms, with reactive triggers on "command not found" errors and proactive mode for environment audits.

Engram is a learning system that turns Claude into a Socratic tutor, verifying your understanding through free-recall challenges and scheduling review sessions using FSRS-4.5 spaced repetition. It keeps you accountable to what you actually learn—with receipts—across nine agentic platforms including Claude Code, OpenAI Codex, OpenCode, and multiple others.