[ Multipliers ]

Multipliers

A living list of multipliers — the reusable artifacts, practices, and methods that raise what one mind can build. Curated, and growing.

20 multipliers

• Agent Skills ↗

  Anthropic's format for packaging instructions, scripts, and resources that an agent loads on demand to do a specific job better; ships an initial open-source set of 17.

  skill LEVERAGE
  link github.com/anthropics/skills more www.anthropic.com/news/skills by @anthropics

• Beads ↗

  A git-backed, dependency-aware issue tracker (CLI `bd`) that acts as persistent memory for AI coding agents, letting them track and sequence long-horizon work across sessions via a queryable task graph.

  tool LEVERAGE
  link github.com/gastownhall/beads more gastownhall.github.io/beads by @gastownhall

• BMAD-Method ↗

  A spec-driven framework that orchestrates named persona-agents (analyst, PM, architect, developer, QA) across a full development lifecycle.

  workflow INTENT LEVERAGE JUDGMENT
  link github.com/bmad-code-org/BMAD-METHOD by @bmad-code-org

• Caveman ↗

  A Claude Code skill that compresses an agent's output into terse 'caveman' prose, cutting 65–75% of output tokens while keeping full technical accuracy — less spend per response without losing substance.

  skill LEVERAGE
  link github.com/JuliusBrussee/caveman more github.com/JuliusBrussee/caveman/blob/main/INSTALL.md by @JuliusBrussee

• Claude Code ↗

  An agentic coding harness that runs a model in a loop over your codebase — reading files, calling tools, executing, and iterating — with subagents for context economy.

  harness LEVERAGE
  link github.com/anthropics/claude-code more docs.claude.com/en/docs/claude-code/overview by @anthropics

• Cursor ↗

  An IDE-integrated agentic coding harness that trades raw autonomy for tight editor context, inline diffs, and a human-in-the-loop review cycle.

  harness LEVERAGE
  link github.com/getcursor/cursor more cursor.com by @getcursor

• Devin ↗

  A cloud-based fully autonomous coding agent that handles long-horizon tasks end-to-end — the harness, tools, and workflow packaged as a service you log into.

  platform LEVERAGE
  link devin.ai by @cognition-ai

• Fabric ↗

  An open-source framework of 230+ reusable prompt "patterns" — each a crafted system prompt stored as markdown that you can pipe into any model from the terminal.

  prompt INTENT LEVERAGE
  link github.com/danielmiessler/Fabric by @danielmiessler

• Gas Town ↗

  A multi-agent orchestration layer for Claude Code and other coding agents that coordinates a fleet of parallel worker agents with git-backed persistent work tracking and crash recovery, via a `gt`/`bd` CLI.

  harness LEVERAGE
  link github.com/gastownhall/gastown more github.com/gastownhall/gastown/tree/main/docs by @gastownhall

• GitHub MCP Connector ↗

  A pre-built MCP server granting agents read/write access to GitHub repositories, pull requests, issues, and CI — one of the most-installed MCP connectors.

  tool LEVERAGE
  link github.com/github/github-mcp-server more modelcontextprotocol.io by @github

• GitHub Spec Kit ↗

  A toolkit for spec-driven development that makes the specification an executable, first-class artifact driving plans, tasks, and code.

  workflow INTENT JUDGMENT
  link github.com/github/spec-kit more github.blog/ai-and-ml/generative-ai/spec-driven-development-with-ai-get-started-with-a-new-open-source-toolkit by @github

• Headroom ↗

  Compresses everything an AI agent reads — tool outputs, logs, files, and RAG chunks — before it reaches the LLM, cutting token usage 60–95% with the same answers, so you get far more work out of each context window and usage cap.

  tool LEVERAGE
  link github.com/chopratejas/headroom more headroom-docs.vercel.app/docs by @chopratejas

• Model Context Protocol ↗

  An open standard giving agents a common way to reach tools and data — thousands of servers with pre-built connectors for GitHub, Slack, Postgres, Google Drive and more.

  tool LEVERAGE
  link modelcontextprotocol.io more github.com/modelcontextprotocol/servers by @modelcontextprotocol

• Open Brain AI ↗

  A computational language-assessment platform that extracts objective linguistic metrics from speech and text, turning natural language into reproducible, evidence-based data instead of subjective evaluation.

  platform JUDGMENT
  link openbrainai.com

• Playwright MCP ↗

  An MCP server that lets coding agents drive a real browser via Playwright's accessibility tree — deterministic web navigation, form-filling, scraping, and end-to-end test automation without screenshots or a vision model.

  tool LEVERAGE
  link github.com/microsoft/playwright-mcp more www.npmjs.com/package/@playwright/mcp by @microsoft

• Ponytail ↗

  A Claude Code skill that makes the agent "think like the laziest senior dev in the room" — walking a restraint ladder before writing, so it produces the minimum code that works.

  skill JUDGMENT
  link github.com/DietrichGebert/ponytail by @DietrichGebert

• Serena ↗

  An open-source MCP server that gives any coding agent IDE-grade, symbol-level code retrieval and editing across 20+ languages via the Language Server Protocol, so agents navigate and refactor large codebases without dumping whole files into context.

  tool LEVERAGE
  link github.com/oraios/serena more oraios.github.io/serena by @oraios

• skills.sh ↗

  An open, cross-agent registry and CLI by Vercel Labs for discovering and one-command installing reusable agent skills across 70+ coding agents (Claude Code, Cursor, Codex, and more).

  platform INTENT LEVERAGE JUDGMENT
  link www.skills.sh more github.com/vercel-labs/skills by @vercel-labs

• Slack MCP Connector ↗

  A pre-built MCP server for agents to search, read, and post across Slack channels and threads, turning a team's message history into agent-accessible context. (Now in the archived reference-servers repo.)

  tool LEVERAGE
  link github.com/modelcontextprotocol/servers-archived/tree/main/src/slack more modelcontextprotocol.io by @modelcontextprotocol

• Work Journal MCP ↗

  A persistent memory tool that gives agents access to decision rationale, session history, and architecture notes across conversations — capturing the why a diff can't.

  tool LEVERAGE JUDGMENT
  link github.com/juliuszfedyk/dev-journal more workjournal.app by @juliuszfedyk