Knowledge work superpowers that compound over time
npx claudepluginhub nabeelhyatt/coworkpowersKnowledge work superpowers that compound over time. Research, execute, review, and capture learnings to make each task easier than the last.
RuFlo Marketplace: Claude Code native agents, swarms, workers, and MCP tools for continuous software engineering
Claude Code marketplace entries for the plugin-safe Antigravity Awesome Skills library and its compatible editorial bundles.
Production-ready workflow orchestration with 79 focused plugins, 184 specialized agents, and 150 skills - optimized for granular installation and minimal token usage
Share bugs, ideas, or general feedback.
Knowledge work superpowers that compound over time.
CoworkPowers is a Claude Code plugin that gives Claude systematic capabilities for tackling knowledge work -- drafting communications, making decisions, preparing for meetings, and more. Each completed task feeds insights back into the system, making the next similar task faster and better.
/coworkpowers:workflow-research) - Thoroughly research the task, search past learnings, gather context/coworkpowers:workflow-work) - Execute the plan with specialized agents/coworkpowers:workflow-review) - Multi-agent quality review from multiple perspectives/coworkpowers:workflow-compound) - Extract patterns, templates, and preferences for next timeResearch runs parallel agents to gather context before any work begins. The context-gatherer pulls relevant background. The stakeholder-mapper identifies who cares and why. The constraint-analyzer finds the boundaries. The precedent-researcher checks what's worked before. These run in parallel -- not sequentially -- so research is fast even when it's thorough.
Stakes classification happens here too. A routine email gets a lightweight context scan. A board presentation gets the full agent roster. This prevents the overwork problem where every task gets treated like a crisis.
Work is where execution happens. The system picks the right agent mindset for the task type: executive-writer for communications, analyst for data work, decision-architect for strategic choices (this one has 40+ decision frameworks it selects from), meeting-orchestrator for meeting prep, coach for leadership challenges. The work uses everything Research gathered, so there's no redundant context-fetching.
Review is where progressive loading gets interesting. Low-stakes work might only get a clarity pass. High-stakes work activates up to 8 specialized reviewers running in parallel:
Each reviewer returns findings tagged by severity: Critical, Important, or Minor. Criticals get fixed before delivery. Important items get flagged. Minor items are noted but not blocking.
After Work and Review complete, the Compound phase runs analysis agents that extract reusable knowledge:
Each insight gets stored as a discrete, tagged file in .context/learnings/. One insight per file, because granular knowledge is findable. A monolithic "things we learned" doc isn't.
Next time you do similar work, the Research phase searches these learnings before doing any new research. It loads matching patterns, applies saved templates, honors your preferences, and avoids documented failures.
The practical effect: your first partner update might take the full Research > Work > Review cycle. Your fourth one loads the template, applies your preferred tone, and skips the research it already has. Faster, cheaper, and more consistent.
# Option A: (Easiest) Open Claude Code, then install from Marketplace just copy and paste this line into Claude Code and hit yes, then restart Claude Code after install and start with the first command - /workflow-research and the first task
/plugin install coworkpowers@coworkpowers
# Option B: Test locally
git clone https://github.com/nabeelhyatt/coworkpowers.git
claude --plugin-dir ./coworkpowers