#403403rd of 455 in Skills, plugins and prompts
most-capable-agent-system-prompt
Long copy-paste system prompt that makes a coding agent build its own agent system
- Stars
- 887
- License
- MIT
- Last commit
- Nov 2026
Missed the latest GitHub check; these facts may be older.
Overview
A roughly 2,950-line (about 23,000-token) prompt you paste into Claude Code, Codex, Cursor or a similar agent. It tells the agent to inspect the workspace, write file-based project state (AGENTS.md, plan.md, tasks.md, knowledge.md, FAILURE.md), and prove one closed loop of goal, task graph, verification and memory update before adding workers, evals and a self-improvement loop. It can wrap an existing agent or build on an agent SDK.
Who it is for: Claude Code, Codex or Cursor users wanting a persistent agent setup
Strengths
- Plain text with no tool-specific syntax; works with any agent that can read and write files
- Requires a verified closed-loop first milestone rather than an architecture essay
- State lives in plain project files, so work can continue from the folder alone
- README documents trimming, customizing, and a terminal copy command
Weaknesses
- About 111,000 characters; Codex's default 32 KiB AGENTS.md limit needs raising to use it as instructions
- Results depend heavily on a strong agentic model; chat-only tools can only plan
- Expect long sessions and real token spend; scope is very broad
- README mixes in unrelated promotional links and paid offers
What it needs
- no GPU
- Needs A coding agent that can read/write files and run commands
- Models: Claude Code, OpenAI Codex, Cursor, Antigravity, OpenClaw
Also in Skills, plugins and prompts
See all 455| Rank | Project | Score |
|---|---|---|
| 1 | claude-memPersistent cross-session memory plugin for Claude Code and other coding agents | 97 out of 100 |
| 2 | impeccableDesign skill with 24 slash commands for AI coding agents' frontend output | 97 out of 100 |
| 3 | gstackMarkdown slash-command skills that give Claude Code planning, review, QA and ship roles | 96 out of 100 |
| #4 | rufloAgent harness adding swarms, memory and hooks to Claude Code and Codex | 96 out of 100 |
| #5 | cavemanShortens coding-agent replies and compresses tool output before the model reads it | 95 out of 100 |