gstack vs Superpowers vs AEGIS — 3 Philosophies of AI Agent Systems

Quick Comparison gstack Superpowers AEGIS Creator Garry Tan (YC President) Jesse Vincent (obra) AEGIS Contributors GitHub Stars ~23K+ (7 days) ~40K+ New (PyPI: aegis-gov) Philosophy Startup sprint workflow Engineering methodology Constitutional governance Approach 15 opinionated skills as roles TDD + debugging + brainstorming framework Boardroom meetings + rule engine + red team Governance None (trust the workflow) Methodology-enforced discipline Explicit rules, verdicts, audit trails Agent Count 6 virtual roles Subagent-driven (dynamic) 9 default + 8 specialists (17 council members) Scalability Solo developer / small team Solo to small team Solo to enterprise (140+ agents in full version) Learning Curve Low — copy skills, run commands Medium — understand methodology first Medium — understand governance model LLM Support Claude Code (+ Codex, Gemini CLI) Claude Code primary Anthropic, OpenAI, Ollama (any LLM) License MIT MIT Apache 2.0 gstack: The Startup Sprint What it is: Garry Tan's