Edgeless Agent Starter Kit: Launch Your AI Swarm on macOS in Under an Hour
The complete starter kit for running a multi-agent AI swarm on macOS. Pre-configured profiles, working cron jobs, 8 starter skills, and the exact setup we use to run 10+ agents daily. No theory. Just configs that work.
Setting up a multi-agent AI swarm from scratch is overwhelming. Hermes Agent has dozens of moving parts: profiles, skills, cron jobs, Paperclip integration, ChromaDB, Discord bridge. Most people give up before they get a single agent working.
This kit is the exact setup we use to run 10+ agents daily on a MacBook Pro. Extracted from 3+ months of production use. Not a tutorial. A working system you deploy in under an hour.
What you get
45-page PDF guide with 6 sections and 5 bonuses:
Section 1: One-Command Setup
setup.shscript: installs Hermes, creates your first profile, connects Discord + Telegram- Tested on macOS 14+ (Apple Silicon)
- Linux instructions included (not automated)
Section 2: Agent Swarm Blueprint
- 5 pre-configured agent roles: research, orchestration, execution, documentation, and audit
- AGENTS.md templates for each role
- Routing matrix: when to use which agent, with real examples
- Paperclip integration config
Section 3: Skill System
- 8 starter skills (research, coding, docs, monitoring, triage, enrichment, sales, dispatch)
- Skill authoring template (YAML frontmatter + markdown body)
- How to load skills on-demand vs. default
- How to add your own skills
Section 4: Automation Stack
- 3 pre-built cron job templates: RSS intelligence, email triage, health checks
- Cron silent-failure detection built-in
- Notification routing: Telegram vs. Discord thresholds
- How to add your own cron jobs
Section 5: Knowledge Base
- Obsidian vault setup (taxonomy, templates, inbox system)
- ChromaDB integration for semantic search
- RSS intelligence pipeline config
- How to query your knowledge base from agents
Section 6: Troubleshooting
- 15+ common failures and fixes from production ops
- Provider fallback patterns (OpenAI to Fireworks to etc.)
- Supervisor threshold tuning
- What to do when an agent goes silent
Bonuses
- Agent Autonomy Policy Template: fill-in-the-blank AGENTS.md for any new agent
- Discord Envelope Protocol Cheatsheet: compact routing reference
- 50 Production Skills Index: curated list of best Hermes skills with install commands
- Weekly Health Check Script: automated system audit
- Paperclip Task Routing Config: project-to-agent mapping template
Who this is for
- AI developers who want to run local agents but the setup is overwhelming
- Solopreneurs who need a research + coding + documentation team that works while they sleep
- AI researchers experimenting with multi-agent systems
- NousResearch community members who want a working Hermes swarm on macOS
Who this is NOT for
- This is macOS-optimized. Windows users need WSL2 or a Linux VM.
- You need API keys for at least one LLM provider (we recommend Fireworks for cost efficiency)
- This is not a beginner's guide to AI. You should know what an LLM is.
Technical requirements
- macOS 14+ (Apple Silicon)
- API key for OpenAI, Fireworks, or similar
- 8GB RAM minimum (16GB recommended for multiple agents)
- ~2GB disk space for the full stack
All configs are tested and working. No "edit this file and hope it works."