Agents

Research & Analysis (11 agents)

• npx agntk - config AI agent CLI with 20+ tools, persistent memory, Ollama auto-detection, and free tier |
• qa-orchestra - agent QA toolkit with 10 specialized agents — orchestrator, environment-manager, functional-reviewer, test-scenario-designer, browser-validator, automation-writer, manual-validator, bug-reporter, release-analyzer, smart-test-selector. Stack-agnostic, output-chained, live validation via Chrome MCP |

Companion Apps & GUIs

GateGuard — Fact-Forcing PreToolUse Gate

• Anima - project ASCII familiars, nim token economy, cross-session watcher daemon. Tauri v2 + Rust, 4MB binary |
• telegram-ai-bridge - bot personas, A2A multi-agent collaboration, tool approval from phone. Supports Claude Code, Codex, and Gemini backends |
• Asynkor - machine sync. One MCP server for Claude Code, Cursor, Windsurf, JetBrains. Open source client, hosted infrastructure |
• ClaudeCode Launchpad CLI - minute installer for Claude Code — auto-installs Node.js, Git & Claude Code; two-line live status bar (model, context %, usage limits with reset timers); desktop shortcut with folder picker; right-click "Open with ClaudeCode Launchpad CLI" on any Windows folder; optional light-blue theme; Claude flags support |
• clideck - like dashboard for managing AI coding agents (Claude Code, Codex, Gemini CLI, OpenCode) in one browser window. Live status, session resume, autopilot routing between agents, mobile remote |
• WhereMyTokens - session token counts, costs, context window progress, rate limit bars (5h/1w), activity heatmaps, and model breakdowns. Registers as a Claude Code statusLine plugin for live data without polling. Windows only |
• Onepilot - | iOS app for running Claude Code, Codex, and other coding agents on remote servers via SSH from your phone. Full terminal emulator with SwiftTerm, agent session management, mobile access to dev machines |
• docker-claude-code - profile support, security hardening, best-practice defaults, a set of pre-installed plugin/skill bundles and remote dev support. Drop-in replacement for claude — a simple shell alias is all it takes. |
• amux - source agent multiplexer for running dozens of parallel Claude Code sessions with web dashboard, self-healing watchdog, kanban board, agent-to-agent REST API, and mobile PWA. Single Python file |
• cc-token-status - machine iCloud sync. 5 languages, dark/light mode. Single Python file, auto-updates |
• Watchfire
• aby-claude-watcher - time macOS dashboard for Claude Code sessions. Five color-coded states (thinking, running, waiting, error, completed), per-session notifications, one-click focus terminal across iTerm2/Terminal/Warp/VSCode/Cursor/Ghostty/kitty/WezTerm/Hyper. Bilingual FR/EN, menu-bar tray, MIT licensed. Apple Silicon + Intel builds |
• Bernstein - driven, deterministic. Apache-2.0 |
• cctally - per-1% trend. Local web dashboard, terminal UI, forecasts, threshold alerts. Apache-2.0, zero telemetry. |
• claude-code-notifier - derived icon (SF Symbols–free), click-through to your IDE (VS Code, Cursor, Windsurf, JetBrains, iTerm2, Terminal) via bundle ID swap, per-event sound labels, hardened inputs (allowlists, length caps, control-character stripping). Universal binary (arm64 + x86_64), macOS 12+, MIT |
• ToutKit - in terminal and an in-app webview that renders whatever the AI writes — each note is a self-contained folder with its own SQLite, key/value store, attached files, and scripts, so a note can grow from a static page into a sortable table, dashboard, or research tool. Local-first, AGPL-3.0 |

Quick Setup with cc-safe-setup

• ctrl - | Pixel-art office that visualizes AI coding agents (Claude Code, Codex, Gemini) working in real time. Standalone web app + daemon, relay sharing for remote access, multi-agent support |
• CloudCLI - source web/mobile UI for Claude Code, Cursor CLI, and Codex. Responsive design, integrated shell, file explorer, git explorer |
• Companion
• Ruflo - deploy multi-agent swarms, coordinate autonomous workflows, distributed swarm intelligence, RAG integration |
• Claude-Code-Workflow - driven multi-agent cadence-team framework with intelligent CLI orchestration (Gemini/Qwen/Codex) |
• ccswarm - based multi-agent orchestration with specialized agent pools, Git worktree isolation, 93% token reduction |
• parallel-code - each in its own git worktree |
• Bernstein - agent orchestrator — spawns Claude Code, Codex CLI, and Gemini CLI in parallel on isolated git worktrees, verifies with tests, auto-commits. Zero LLM tokens on coordination |
• Poirot - running commands. Reads local transcripts, runs offline |
• TokenEater
• Claw
• The Claude Protocol - blocks unsafe operations, enforces worktree isolation |
• crit - hook to review and approve plans before execution, outputs structured .crit.json for agent consumption. |
• Untether
• OctoAlly - multi-agent hive-mind via RuFlo, Whisper voice dictation, tmux session persistence, built-in git source control, Electron desktop app |
• tokburn - calculates API-equivalent costs, detects waste patterns (repeated reads, floundering, cost outliers), serves web UI + REST API. Zero network calls, zero accounts, zero telemetry. uvx tokburn serve. |
• skill-builder - in skills + unlimited AI-generated. React UI with MCP server builder |
• claude-workspace-snapshot - coded Windows Terminal tabs. Detects running sessions via process inspection and .jsonl file activity. Windows only |
• KANBAII - native kanban board for Claude Code — plan visually, track progress, let AI execute. Sequential engine (Ralph), parallel multi-agent (Teams), cost tracking, real-time dashboard. npx kanbaii start |
• Claude Session Visualizer
• CCHub - MCP servers, skills, hooks, config profiles. Auto-scans Claude Code, Claude Desktop, Cursor configs |
• AionUI - source 24/7 Cowork app for Gemini CLI, Claude Code, Codex, and OpenCode |
• chops

Ecosystem

GateGuard — Fact-Forcing PreToolUse Gate

• claude-token-lens - time token attribution from local Claude Code JSONL sessions — see which tool, agent, MCP server, or skill is burning your quota. Live Ink dashboard, burn rate, ETA, 5h + 7-day tracking, zero telemetry. npm install -g claude-token-lens |
• cc-discipline - breaker hard-stops at 5 edits, pre-edit-guard enforces debugging process, 7 rules, 7 skills, 2 subagents. Interactive installer with append mode. bash ~/.cc-discipline/init.sh |
• claude-memory-bridge - | Cross-project memory sharing via namespaces. Adds a shared layer between global and project-scoped memory so projects selectively share knowledge. MCP server, no database |
• claude-overlay - | CLI for managing Claude Code project configs across custom providers (Databricks, Bedrock, OpenRouter, LiteLLM, Cloudflare). Overlay merge/remove, MCP web search setup, multi-provider switching, team config export/import |
• faf-cli - | The package.json for AI context. IANA-registered .faf format — init, score, bi-sync with CLAUDE.md, export to AGENTS.md, GEMINI.md, .cursorrules. Built with Bun. |
• git-parsec
• Claude Code AWS Gateway - hosted API gateway that routes Claude Code through Amazon Bedrock with team API keys, budgets, rate limits, OIDC SSO, SCIM provisioning, and an admin portal for analytics |
• cc-agents-md - session reload, context preservation, custom patterns, and stat-based caching. GPL-3.0 |
• cognitropy-lab - | Autonomous workspace factory powered by a date-seeded entropy engine (363 domains, 22 categories, 18.8M unique outcomes). Builds one new Agent Workspace Model-compliant workspace daily — from firmware RE to aquaponics to satellite comms — with full slash commands, workflows, and secrets scanning. Zero human intervention.
• Wiggum CLI - | Open-source AI agent that scans codebases (80+ tech), generates specs through AI interviews, and runs autonomous Ralph loops via Claude Code or Codex. Agent mode ships GitHub issues end-to-end with priority-aware scheduling and auto-merge |
• expert-dispatch - let cheap AI assistants delegate complex tasks to Claude Code CLI. Bash adapter with run/resume/review workflow, session continuity, per-project directories, and structured JSON output. Designed to be called by another AI system |
• PRISM - format CLAUDE.md recommendations. Read-only, zero telemetry |
• Not Human Search MCP - | Agent-first search engine for AI-accessible tools — 9,000+ sites indexed. Search by category, check agentic scores, verify MCP endpoints. MCP server at nothumansearch.ai/mcp, REST API at /api/v1/search. Go. |
• AI Dev Jobs MCP - | AI/ML job board with 5,200+ roles across 260+ companies. Search jobs, filter by tags/salary/location, post listings programmatically. MCP server at aidevboard.com/mcp, REST API at /api/v1/jobs. Go. |
• codachi - style statusline pet that grows with context usage, shows cache hit rate and burn speed |
• claude-code-memory-guide - to-production memory setup: CLAUDE.md + MEMORY.md + MemPalace semantic search + session hooks + warm-model daemon + nightly Karpathy compile with free-LLM key rotation. Three guides — linear tutorial, memory format reference, automation reference. Runnable scripts, secrets-in-git precommit guard, end-to-end verification checklist. MIT |
• cc-hud - line Claude Code statusline -- model, context usage (1/8-char progress bar), active subagents, 5h/7d rate-limit countdowns. Pure Node.js, zero deps, Windows-safe. |
• Claudoscope - text search, real-time secret detection, config health linting (44 rules). Reads local JSONL files, fully offline, zero telemetry. Swift/SwiftUI, Homebrew install |
• LynxPrompt - source platform for managing AI coding agent configs (CLAUDE.md, AGENTS.md, .cursorrules, copilot-instructions.md). Web marketplace, CLI, VS Code extension, self-hostable via Docker/Helm with federation support |
• Solo Orchestrator - gated development methodology for building production-quality applications with Claude Code. One-command setup generates CLAUDE.md, CI/CD pipelines, pre-commit hooks, and security tooling. Auto-discovers new platforms and languages from the filesystem. Includes enterprise governance, intake wizard, and adversarial evaluation prompts. Example project shows the complete artifact trail. |
• vibe-replay - replay.com) |
• AgentsInFlow - hosted desktop workspace for governed AI development — run Claude Code, Codex, Cursor, OpenCode in isolated runtimes with persistent memory, ticket-driven orchestration, git worktrees per execution, browser bridge, 28 diagram types, session forking. Free early access, no sign-up. Electron (macOS/Windows) |
• SpecLock - commit law. Catches euphemisms, temporal evasion, synonym substitution, compound violations. 51 MCP tools, 1,009 tests, on Official MCP Registry. npx speclock protect |
• TokRepo - | Curated, community-ranked registry of 600+ agent skills, MCP servers, prompts, scripts, workflows, and configs. Cross-platform tagging (Claude Code, Codex CLI, Gemini CLI, Cursor), upvotes and usage data |
• claude-skins - command install, pure bash engine. |
• Agent Sessions
• Claude Code Skills 中文精选集 - | Chinese curated directory of 140+ Claude Code Skills, Agents, Plugins, and workflows for Chinese-speaking Claude Code users |
• nylas/cli - in MCP server with 16 tools across Gmail, Outlook, Exchange, Yahoo, iCloud, and IMAP. One-line install: nylas mcp install. Works with Claude Code, Cursor, Codex, and Windsurf. Docs: https://cli.nylas.com |
• agenttrace
• agent-shadow-brain - powered background code analysis agent for Claude Code that continuously monitors your codebase and provides intelligent suggestions. npm: @theihtisham/agent-shadow-brain |
• omni-skills-forge - command manager for AI coding assistants. Create, share, and manage reusable skills across tools. npm: omni-skills-forge |
• cc-tempo - clock active work time, SubAgent speedup ratio, /clear-resilient PID timer, code-churn sparkline with trend arrow, and multi-instance detection. Bash + TypeScript, MIT. |
• The Froject - based wizard that generates complete Claude Code workspaces for go-to-market teams (marketing, sales, customer success, HR, finance, operations). Downloads a ZIP including CLAUDE.md, context files, and pre-built skills, commands, agents, rules, hooks, and settings tailored to your role. Client-side, no backend. Plugin marketplace at thefroject-plugins. |
• Bring Your AI - | Local-first Claude Code to Codex migration path. Keeps harness files local, maps CLAUDE.md/AGENTS.md guidance and MCP config, and adds a Codex import checklist for hooks, secret refs, and non-equivalent behavior before Codex edits code. |
• Drevon
• systemprompt-template - | Governance infrastructure for Claude Code. Single compiled Rust binary that authenticates, authorises, rate-limits, logs, and attributes costs for every AI interaction before it reaches a tool or database. Self-hosted, air-gap capable, MCP + A2A compatible. BSL-1.1 |
• llm-prices - prices` |
• claude-starter-kit - to-use project structure with persistent memory, session continuity, hooks, and 3 bundled skills (Gemini, Brainstorm, Design) |

Quick Setup with cc-safe-setup

• cc-statistics - in-one Claude Code stats: CLI + Web + native macOS SwiftUI panel. Token costs, code changes by language, efficiency scoring, weekly reports. Supports Codex and Cursor too |
• cog - persistent memory, self-reflection, and foresight via plain-text conventions. Zero dependencies, just CLAUDE.md + markdown files |
• Cortex - | Persistent AI memory for coding assistants. Auto-captures decisions, patterns, and context across sessions. VSCode extension + CLI + MCP server. Free |
• openclaw-self-healing - tier autonomous crash recovery for Claude Code and any service — 64% auto-resolved, LLM-agnostic (Claude/GPT-4/Gemini/Ollama), Prometheus metrics |
• openclaw-memorybox
• openclaw-self-evolving - improvement pipeline — scans Claude Code logs, proposes CLAUDE.md/AGENTS.md rule changes, zero API cost |
• caliber - | CLI that fingerprints projects and generates AI agent configs (CLAUDE.md, skills, AGENTS.md). Scores quality, auto-refreshes, supports Claude Code + Cursor + Codex |
• claude-code-kickstart - detect, 12 agents, 10 profiles, and 20+ shell commands |
• claude-code-power-stack - style planning in a single install with cheat sheet PDF |
• clooks - invocation spawning -- 112x faster hooks with batching, dependency resolution, metrics |
• AIRIS MCP Gateway - based MCP multiplexer that aggregates 60+ tools behind 7 meta-tools, reducing context token usage by 97%. One command to start, auto-enables servers on demand |
• gemini-claude-bridge - to-Claude protocol converter for using Gemini models as Claude Code backend. Fixes 3 LiteLLM bugs |
• AgenTopology - agent orchestration -- define agents, flows, gates, hooks, group chats in one .at file, scaffold to 7 platforms, interactive visualizer. Apache 2.0 |
• brood-box - isolated microVMs with snapshot isolation, egress control, and MCP authorization |
• claude-kit - 13 composable stacks, 6 agents, audit scoring, practices pipeline. Bootstraps and maintains .claude/ across projects |
• GAAI Framework - in governance layer (.gaai/ folder) -- backlog authorization, cross-session memory, decision tracking, QA gates, autonomous delivery daemon. Claude Code + Cursor + Codex CLI + Gemini CLI |
• Claude Command Center - | Production-tested workspace template with session hooks, structured memory (4 types), multi-domain routing, daily task tracking, and interactive /setup onboarding. 6 domain templates, pre-commit safety, example skills |
• Code Insights - first CLI and dashboard for analyzing AI coding sessions from Claude Code, Cursor, Codex CLI, Copilot CLI, and VS Code Copilot. SQLite-backed with terminal analytics, browser dashboard, and LLM-powered insights |
• claude-code-cheat-sheet - page printable reference — all shortcuts, commands, CLI flags, MCP config, memory, skills, agents. Auto-updated daily from official docs. EN · ZH · JA · KO |
• agent-dotfiles
• zclean
• Hindsight - | State-of-the-art long-term memory for AI agents by Vectorize. Biomimetic retain/recall/reflect with 4 parallel retrieval strategies. Self-hosted or cloud, MIT-licensed. Claude Code integration |
• healthcare-agents - level expertise across 10 divisions -- revenue cycle, compliance, quality, clinical ops, payer relations, health IT, and more |
• cc-switch - platform desktop All-in-One assistant for Claude Code, Codex, OpenCode, and Gemini CLI |
• claude-code-router
• SuperClaude
• claude-code-templates
• planning-with-files - style persistent markdown planning skill for structured project management |
• claude-hud
• awesome-claude-code-subagents
• serena - code-aware search and navigation |
• Understand-Anything
• n8n-mcp

Hooks

GateGuard — Fact-Forcing PreToolUse Gate

• GateGuard - ai/) for details. By ZUNO WORKS K.K.

Hook Scripts

• `smart-approve.py` - commands and check each against allow/deny patterns |

Related SDKs

• claude-code-hooks - testable .handle() method. Zero dependencies.
• EchoCoding - triggered SFX, ambient soundscape, and optional cloud TTS/ASR voice interaction. Works with Claude Code hooks and also supports Cursor/Windsurf, Codex CLI, and Gemini CLI.
• claude-code-hooks - testable .handle() method. Zero dependencies.

Plugins

All Plugins

• skills-janitor
• claude-agentic-coding-playbook - based practices for LLM-assisted development -- hooks, skills, scripts, and a best-practices guide with 58 citations. Includes 19+ guard/lifecycle hooks, investigation workflow, fleet indexing, and claude-loop for autonomous task queues. |
• claude-code-mcp - shot MCP server -- an agent in your agent. Permissions bypassed automatically. By Peter Steinberger. 1,100+ stars |
• claude-code-system-prompts - 18 builtin tool descriptions, sub-agent prompts, utility prompts. Updated for each release. 5,900+ stars |
• claude-context
• claude-cost-optimizer - 60% on Claude Code costs -- 6 deep-dive guides, 15 ranked strategies, 12 CLAUDE.md templates, token estimator, usage analyzer, and copy-paste configs |
• fractal - evaluates as it learns |
• Bouncer - liner install. |
• ccmanager - approval via Haiku, devcontainer support. 940+ stars |
• ccpm - analyze, epic-start, epic-merge commands. 7,600+ stars |
• chief
• claude-hooks-mastery - UV single-file Python scripts, sub-agents, meta-agent, team-based validation, AI-generated audio feedback. 3,300+ stars |
• claude-mem - text search. 35,900+ stars |
• claude-notifications-go - platform smart notifications -- 6 types, click-to-focus, context analysis, webhooks. Single Go binary, zero deps. 340+ stars |
• claude-scaffold - activate via hooks based on project context. Cross-repo sync via update --all. Install: npx claude-scaffold init |
• claude-recap - topic session memory using Shell hooks — archives each conversation topic as a separate Markdown summary. Two hooks, bash + Node.js, 100% local |
• claude-rank - fixes robots.txt, sitemap.xml, llms.txt, and JSON-LD. 170+ rules across 10 scanners, zero config |
• claude-sounds
• claude-code-hooks - tested hooks from 160+ hours of autonomous operation. Destructive command blocker, branch guard, syntax check, context monitor, permission auto-approver. Bash, zero deps |
• cc-safe-setup - safe-setup`) to install 6 essential safety hooks in 10 seconds. Zero dependencies |
• claude-supermemory
• eliniscan - fixes. |
• cozempic - updating now, atomic writes, strict session guard, zero false positives on team detection. 13 pruning strategies, Agent Team protection, MCP server, JSONL doctor. Install: /plugin marketplace add Ruya-AI/cozempic |
• cup
• claw-army/claude-node - json |
• discoclaw - based automation |
• jarvis - layer self-healing infrastructure. Uses claude -p headless mode at $0 extra cost. |
• jarvis-company-board - time AI agent collaboration board built on Next.js 15 and SQLite WAL — 8 named AI board members debate decisions via SSE push, with a DEV task approval workflow and Railway deploy support. |
• dna-claude-analysis - style HTML dashboard |
• codetape - records semantic traces and syncs README, CHANGELOG, CLAUDE.md. Zero deps. npx codetape init |
• faf-skills
• lightcms - native CMS with 41 MCP tools for managing websites through natural language — pages, templates, assets, themes, collections, redirects, and more with full content versioning |
• myclaude - agent orchestration routing tasks to Claude Code, Codex, Gemini, and OpenCode based on complexity. OmO skill for intelligent routing. 2,400+ stars |
• oh-my-claudecode - first multi-agent orchestration. 19 specialized agents, 28 skills. Full autonomous execution, Socratic questioning, N coordinated agents. 9,900+ stars |
• onWatch - source Go CLI that tracks AI API quota usage across 7 providers (Synthetic, Z.ai, Anthropic, Codex, GitHub Copilot, MiniMax, Antigravity) with a background daemon (<50 MB RAM), zero telemetry, and a Material Design 3 web dashboard |
• obey - specific, project-local), active blocking via PreToolUse, completion checklists via Stop hook, audit trail, auto-detects rule-like language. Cross-platform. |
• opcode
• oss-autopilot - autopilot` |
• peon-ping - detects SSH/devcontainers. 3,900+ stars |
• preflight - tool MCP server that catches vague prompts before they cost 2-3x in wrong→fix cycles. 12-category scorecards, correction pattern learning, cross-service contract awareness, session history search with LanceDB vectors, and cost estimation. npx preflight-dev |
• production-grade - agent autonomous pipeline — PM, Architect, Backend, Frontend, QA, Security, Code Review, DevOps, SRE, Data Scientist, Technical Writer, Skill Maker, Polymath co-pilot. Two-wave parallel execution, brownfield-safe. |
• pulse
• PUIUX Pilot - configures Claude Code hooks, MCPs, and skills for any project. Scans 95+ project types, selects from 28+ hooks, scores quality (0-100), translates configs across AI tools. Safe: dry-run, atomic writes, backup + rollback. |
• pulser - classifies, prescribes fixes with templates, --fix with rollback |
• reporecall - sitter AST indexing (22 languages), hybrid keyword + vector search, call-graph traversal, hooks + MCP. Injects relevant context in ~5ms before Claude starts thinking. npm: @proofofwork-agency/reporecall |
• the-pragmatic-pm
• vibe-kanban - based orchestration for 10+ coding agents (Claude Code, Codex, Gemini CLI, Copilot, Amp). Isolated git worktrees per agent, inline diff review. 23,200+ stars |
• VibeGuard - time interception and static scanning across 5 languages |
• wshobson/agents - agent workflow orchestrators, 146 skills, 79 tools in 72 focused plugins. 31,300+ stars |
• background-timer - set delayed checks without blocking conversation |
• notch-so-good - art crab (Chawd) lives in your Mac's notch and watches Claude Code for you. Live session timers, color-coded notifications, 13 idle animations, mouse-reactive eyes, drowsiness system. Universal binary, one-line install: npx notch-so-good. MIT, 130+ users |
• agento-patronum
• great_cto - lead, senior-dev, qa-engineer, security-officer, devops, l3-support, project-auditor), 12-angle code review, 10 project archetypes, 13 compliance frameworks (SOC2/HIPAA/PCI-DSS/GDPR/ISO 27001), two-gate approval flow. Opus 4.7 advisor escalation, file-based, MIT |
• AgentLint - backed checks across 5 dimensions. Claude Code plugin. |
• brooks-lint
• getburnd - first cost-control CLI for Claude Code. Reads ~/.claude/projects/*.jsonl, identifies 8 leak patterns (verbose context, tool loops, large file re-reads), prints savings estimates, generates a shareable report URL. MIT, zero telemetry. |
• building-multiagent-systems - agent systems: orchestrator-worker, pipeline, debate, and MapReduce topologies |
• claudebase - machine conflict detection, automatic backups. 158 tests |
• claude-ops
• clirank-mcp-server - mcp-server)) | API discovery for agents -- scores 210+ APIs on CLI-friendliness across 11 signals. Find the right API for any task via MCP tools or REST (curl "https://clirank.dev/api/discover?q=send+emails"). No auth required |
• /cht - proof auto-save via hooks, project-based organization, search, and management. npm: @kosoukhov/cht-cli |
• claude-code-sessions - text search, token analytics, task management, session comparison, timeline views, and context recovery across all sessions. Includes a web dashboard. Zero runtime dependencies, read-only |
• cc-aws-keepalive - proactive expiration warnings, credential refresh bypass for Bedrock SSO/SAML, optional statusline countdown timer. 4 hook scripts, zero dependencies |
• knowledge-graph - dependency, git-native memory layer for Claude Code that persists learned context across sessions. Pure bash, ~3ms/event, privacy-first |
• cortex - backed retrieval with thermodynamic decay, backed by 41 published papers. PostgreSQL + pgvector + sentence-transformers. 6 lifecycle hooks (SessionStart, UserPromptSubmit, PostToolUse, SessionEnd, Notification, SubagentStart), hierarchical recall, causal chains, knowledge graph navigation. Install and forget. |
• codesight
• dig2crawl - powered CSS selector discovery. 4-level AI extraction escalation (CSS, browser actions, Claude Vision, captcha). Rust. |
• gate4agent
• harness-evolver - native autonomous agent optimization. Multi-agent proposers evolve prompts, routing, tools, and architecture in isolated git worktrees. Install: npx harness-evolver@latest |
• idle-timing - turn duration) into every user prompt so the model knows how long the conversation has been paused. Ships a statusline fragment for a live elapsed-time readout and a visible [after Xm Ys] note when you return from >10s idle. 57 tests, dual Unlicense/CC0. |
• immich-photo-manager - hosted Immich photo library into a conversation — natural language search, geographic album curation, duplicate detection via perceptual hashing, library health audits, and interactive HTML galleries. 22 MCP tools, 11 skills, 5 slash commands |
• leapfrog-mcp - session browser MCP server -- 15 parallel isolated Chromium sessions so multiple Claude Code terminals can browse simultaneously. 37 tools for navigation, extraction, interaction, network interception, and session profiles. 797 tests. Install: npx -y leapfrog-mcp |
• nexus-agents - native /agents + hooks. Install: /plugin marketplace add williamzujkowski/nexus-agents |
• ORCH - retry, inter-agent messaging, goals, and TUI dashboard |
• paco-framework - first multi-agent OS for Claude Code. Coordinates 3-16 specialized AI agents (Engineering, QA, Growth, Finance) with file-based dispatch, institutional memory, CEO Gate approval, and 24/7 scheduling. No Python required. MIT |
• product-org-os - based agents (CPO, VP Product, PM, PMM, BizOps, CI, and more) modeling a full product org. Two gateways: /product auto-routes to the right agent, /plt runs a multi-stakeholder leadership meeting. MIT. claude plugins install github:yohayetsion/product-org-os |
• review-squad - perspective code review via dispatched subagent panels: experts, normie users, pedantic nitpickers, and real-user task runners |
• sitemd
• simmer - defined criteria across multiple rounds |
• test-kitchen
• whatsapp-claude-plugin - connects as a linked device via Baileys v7 with bidirectional messaging, full media support, voice transcription, permission relay, and access control |
• codebase-graph - language tree-sitter AST parsing, FalkorDB knowledge graphs, 0.944 MRR search quality. npm: @anthropic/codegraph |