Softaworks Agent Toolkit
43-skill curated toolkit for writing, docs, planning, and workflows
About
Softaworks Agent Toolkit is a curated MIT-licensed collection of 43 skills plus agents and slash commands for AI coding agents. The productivity and writing standouts include writing-clearly-and-concisely, professional-communication, humanizer, daily-meeting-update, session-handoff, and marp-slide, alongside documentation skills (READMEs, C4 diagrams, Mermaid) and planning skills (requirements-clarity, ship-learn-next). Every skill pairs a SKILL.md with a user-facing README and follows the Agent Skills format. Install with npx skills add softaworks/agent-toolkit for Claude Code, Codex, Cursor, and others, or register it as a Claude Code plugin marketplace. With 2,500+ stars and 90 commits, it is one of the better-vetted third-party packs. Limits: breadth means uneven depth across skills; some entries are thin wrappers around common advice.
What it can do
- 43 skills across writing, docs, planning, and utilities
- Writing and professional-communication skills
- Session handoff and daily standup workflows
- Marp slide and diagram generation
- Slash commands and specialized sub-agents
- MIT licensed, agent-format compliant
Setup
Run npx skills add softaworks/agent-toolkit and pick your agents, or register /plugin marketplace add softaworks/agent-toolkit in Claude Code and install individual skills like humanizer@agent-toolkit.
Prerequisites
Node.js for the Skills CLI, or Claude Code for the plugin route.
Use cases
Works with
Trust score — 8/10
Scored 8/10 (verified tier, metrics-v1). Evidence: Live GitHub page read this session: 2,548 stars, MIT license, 90 commits; install command copied verbatim from README.; 43 skills verified in README table, including writing-clearly-and-concisely, professional-communication, humanizer, daily-meeting-update, session-handoff, marp-slide.; Breadth-over-depth caveat noted: quality varies across the 43 skills..
★ 2,548 GitHub stars
Sources
Every fact above was checked against these sources. Where a claim comes from the provider and couldn't be independently verified, the text says so.