A complete AI-native software engineering operating system: 7 domain skills, a 20-phase production lifecycle orchestrator, and deep wiki knowledge for building production-grade AI systems.
概览
A cognitive infrastructure system for production-grade software engineering — powered by Hermes Agent. Not a chatbot wrapper. Not a tutorial collection. A full operating system for how an AI agent drives you through the real complexity of building production systems — from the first architectural decision to continuous learning in production. Most engineers learn frameworks. Few learn the underlying engineering discipline. When you build a real system — distributed, secure, observable, AI-native — you're not just writing code. You're navigating a web of decisions across multiple domains simultaneously. Mess up the threat model at Phase 0, and no amount of circuit breakers saves you at Phase 12. This kit gives your AI agent the structured knowledge to guide that navigation. Hundreds of concepts extracted from canonical engineering texts, organized into seven domains, cross-linked by the master orchestrator: Each node in that graph is a standalone concept page.
README
以下为原文 README 节选，完整内容请查看原文。
agentic-swe-kit
A cognitive infrastructure system for production-grade software engineering — powered by Hermes Agent.
Not a chatbot wrapper. Not a tutorial collection. A full operating system for how an AI agent drives you through the real complexity of building production systems — from the first architectural decision to continuous learning in production.
What Problem Does This Solve?
Most engineers learn frameworks. Few learn the underlying engineering discipline.
When you build a real system — distributed, secure, observable, AI-native — you’re not just writing code. You’re navigating a web of decisions across multiple domains simultaneously. Mess up the threat model at Phase 0, and no amount of circuit breakers saves you at Phase 12.
This kit gives your AI agent the structured knowledge to guide that navigation.
The Knowledge Graph
Hundreds of concepts extracted from canonical engineering texts, organized into seven domains, cross-linked by the master orchestrator:
Each node in that graph is a standalone concept page. Each cluster is a domain skill. The master orchestrator knows the connections between clusters.
What’s Inside
7 Domain Skills
Routing layers over deep engineering knowledge. Each skill knows which concepts to invoke and when.
| Skill | Domain |
|---|---|
engineering-mindset |
Quality decisions, trade-offs, pragmatism, estimating |
modular-architecture |
Boundaries, dependency direction, SOLID, clean layers |
production-readiness |
Stability patterns, timeouts, circuit breakers, operability |
distributed-systems |
Consistency, fault tolerance, naming, coordination |
security-engineering |
Threat modeling, crypto, access control, privacy by design |
data-systems-engineering |
Storage, replication, partitioning, transactions, streaming |
llmops-ai-agents |
Agent architecture, RAG, evaluation, observability, guardrails |
1 Master Orchestrator
agentic-swe-master — the system that ties everything together:
- 20-phase production lifecycle from cognitive design through continuous learning
- Diagnostic protocol — asks 5 questions and tells you which skills to load
- Fast routing table — 17 known cross-domain problem patterns, instant routing
- 9 cross-domain rules that apply regardless of domain
- 7 anti-patterns with explicit fixes for each
Wiki (Concept Pages)
Hundreds of full concept pages across 7 domains. Not summaries. Full extractions — definition, mechanism, when to use, trade-offs, worked examples. The skill points you here when you need depth.
Quick Install
Requirements: Hermes Agent installed and configured.
git clone https://github.com/ayush488-glitch/agentic-swe-kit
cd agentic-swe-kit
chmod +x install.sh
./install.sh
Then restart your shell (or run source ~/.zshrc / source ~/.bashrc).
推荐工具
换一个关键词，或者移除筛选条件。
安装
npx skillfish add ayush488-glitch/agentic-swe-kit