# WorldFlowAI/everything-claude-code

Source: [GitHub](https://github.com/WorldFlowAI/everything-claude-code)  
Feed7 permalink: https://feed7.dev/p/everything-claude-code-11d9320  
Published: Unknown  
Trust: Needs Review (needs_review)

## Why Included

A reusable Claude Code harness bundles subagents, skills, hooks, rules, evals, and context persistence. Treat it as a pattern library to adapt, not a configuration to enable wholesale.

## Source Summary

The repository packages **specialized subagents**, workflow skills, slash commands, modular rules, MCP configs, and hooks for saving context and extracting reusable patterns. Its Node.js scripts support **Windows, macOS, and Linux**.

## Practical Implication

Borrow individual pieces that address a measured weakness in your agent loop: verification commands, session checkpoints, retrieval patterns, or scoped reviewers. Keep the harness small enough that its rules and automation remain legible.

## Agent-Ready Context

The repository packages **specialized subagents**, workflow skills, slash commands, modular rules, MCP configs, and hooks for saving context and extracting reusable patterns. Its Node.js scripts support **Windows, macOS, and Linux**.

Borrow individual pieces that address a measured weakness in your agent loop: verification commands, session checkpoints, retrieval patterns, or scoped reviewers. Keep the harness small enough that its rules and automation remain legible.

The configurations reflect one author's workflow, and the repository warns that enabling too many MCP servers can reduce an advertised **200k context window to 70k**. Its recommendation to keep **under 10 MCP servers** is practical guidance, not a demonstrated universal threshold.

## Connected Context

Feed7 judgment across 693 accumulated Signals:

This is a concrete, cross-platform harness kit for selectively adding subagents, skills, checkpoints, hooks, and reusable learning to Claude Code. It reinforces the value of encoded workflows, but also makes configuration load a practical constraint: its own context-window warning supports measured, minimal adoption rather than installing the entire catalog as a default.

- [affaan-m/ECC](https://feed7.dev/p/ecc-0438yju) — ECC expands the same harness-building direction across multiple clients and more control surfaces, reinforcing the need for selective installation, auditing, and pruning.
- [EveryInc/compound-engineering-plugin](https://feed7.dev/p/compound-engineering-plugin-06exf60) — Compound Engineering offers a narrower six-stage lifecycle for persisting plans, reviews, and lessons, providing an opinionated alternative to assembling a broad toolkit piecemeal.
- [Agents Are Where Microservices Were in 2015 — Roberto Milev & Uday Kanagala, Navan](https://feed7.dev/p/agents-are-where-microservices-were-in-2015-roberto-milev-uday-kanagala-1pk4hle) — Navan’s single-agent-first approach supplies an adoption constraint: establish testable skills, persistence, traces, and guardrails before using the repository’s multi-agent features.
- [How our agents build on-brand pages with design.md](https://feed7.dev/p/how-our-agents-build-on-brand-pages-with-design-md-1gt2ft4) — Vercel provides concrete evidence for one harness pattern here: combine agent-readable rules with reusable primitives and repeatable evaluation instead of relying on prose prompts alone.

## Context Map

- Layer: agent
- Domains: coding
- Topics: harness-engineering, skills, subagents

## Uncertainty

- The configurations reflect one author's workflow, and the repository warns that enabling too many MCP servers can reduce an advertised **200k context window to 70k**. Its recommendation to keep **under 10 MCP servers** is practical guidance, not a demonstrated universal threshold.

## Agent Instruction

Use this item as source-backed context. Do not invent claims beyond the linked source. If this item conflicts with another source, call out the conflict.