Find plugins

by claude-contrib

macOS Notification Center alerts for Claude Code — notifies when Claude needs your input

by voitta-ai

Confine agent-run shell commands to one tree with macOS sandbox-exec: reads, writes, symlinks, runtime paths.

by voitta-ai

Fix macOS Sparkle updater error 4005 / 'Failed to create installation cache directory' that persists after stripping quarantine:

Maltese POS 工作區插件. Includes 6 skills.

by Rajiv Ayyangar

Master of Panes — orchestrate parallel Claude Code dev sessions across tmux panes with state tracking, auto-cleanup, and slash commands.

mattpocock-skills is an open-source Claude Code plugin indexed from GitHub. Review its manifest and source before installing.

mattpocock-skills is an open-source Claude Code plugin indexed from GitHub. Review its manifest and source before installing.

mattpocock-skills is an open-source Claude Code plugin indexed from GitHub. Review its manifest and source before installing.

mattpocock-skills is an open-source Claude Code plugin indexed from GitHub. Review its manifest and source before installing.

mattpocock-skills is an open-source Claude Code plugin indexed from GitHub. Review its manifest and source before installing.

by voitta-ai

Diagnose and clear a system-wide file-descriptor exhaustion on a macOS workstation caused by orphaned `mcp-language-server` (LSP-to-MCP bridge) processes leaking file descriptors until the kernel file table is full. The symptom masquerades as whatever tool happens to open a file next - "ENFILE: file table overflow" from a CLI, terraform failing, DNS lookups timing out - so it reads as that tool's bug, not the machine's. Use when: (1) any process on a dev machine dies with `ENFILE: file table overflow` or errno 23, (2) unrelated tools start failing at once with open/socket errors while the machine otherwise seems fine, (3) `sysctl kern.num_files` is within a few percent of `kern.maxfiles`, (4) `ps` shows many `mcp-language-server` (or other per-session MCP bridge) processes with PPID 1. Verified 2026-08-19: 20 orphaned bridge processes held 255,298 fds (kern.num_files 275,146 of kern.maxfiles 276,480 - 99.5%); killing them dropped usage to 19,885 and the failing tool ran clean.

by Divyekant Gupta

Memories discipline plus a portable Codex setup guide; the published npm installer owns hooks and MCP wiring.

by Divyekant Gupta

Self-hosted semantic memory with graph-aware search (PPR multi-hop traversal), temporal reasoning (version preservation, date-range filters), multi-backend routing, 6-signal hybrid search, auto-feature metrics, operator workbench, lifecycle policies, and full audit trail

by chock-core

trigger: repeated mistakes, rediscovered patterns, preferences, non-derivable facts. avoid: persisting file contents, git history, or task intermediates as memory.

by chock-core

trigger: repeated mistakes, rediscovered patterns, preferences, non-derivable facts. avoid: persisting file contents, git history, or task intermediates as memory.

by MattsomStudio

Provenance-aware memory for Claude Code: a generated MEMORY.md index, per-prompt retrieval of un-loaded memories, and the memsom MCP tools (retrieve, ask, blame, revoke...). Requires `pip install memsom` so the `memsom` / `memsom-mcp` executables are on PATH.

by voitta-ai

Audit a multi-source metrics client before believing its zeros: a 0 usually means the source never populates the field, not that the user did not do it.

by wasikarn

A Claude Code harness that composes mattpocock-skills instead of duplicating it. Nine computational deny/ask gates (irrecoverable commands, subagent git guard, subagent spawn guard, task-complete separation, test integrity, config-write guard, secret-content scan, codex-setup guard, routine-trigger guard) plus a SubagentStop verdict-check gate, a 4 KB METHODOLOGY injected at SessionStart, and memory-lint for the auto-memory store.

by voitta-ai

A Micronaut management endpoint 401s with no security library on the classpath, because endpoints.* is read from the YAML document root and a copy nested under micronaut: binds to nothing.

by Abeansits

On-demand consultation with Codex, Gemini, Antigravity, OpenCode, or Oz at any point during development, including video analysis

by ncaq

Respond to PR review comments semi-automatically. Splits review findings into logical units, fixes and commits them one by one, then replies and resolves review threads.

by voitta-ai

Run a multi-phase feature as stacked worktree PRs so each phase reviews independently.

by ActionsCommunity

Operate Multirunner through focused setup, host, diagnosis, GitHub integration, and Windows QEMU skill routers.

by Jerry Boonstra

Reads the plugin. Writes the verdict. Edits nothing. Before you install someone else's plugin or skill pack, neckbeard inventories what it injects and when, whether it reaches your subagents, and what state it persists, then diffs every rule it carries against the rules you already run and sorts each one into duplicate, conflict, or genuinely new. Zero hooks and zero agents; ~270 tokens always-on (the skill description) and ~4.7k when invoked.