I will setup and fix hermes agent multi agent hermes desktop, vps telegram, claude mcp


About this gig
Is your Hermes Agent replying to everyone instead of just you? Losing chat history, gateway silent, VPS install failing, or desktop app crashing? I fix and build Hermes Agent end to end.
I handle Hermes agent setup and Hermes agent fix across desktop, VPS, and gateways: Hermes desktop app install, Hermes vps setup, Hermes telegram bot and Hermes discord setup with owner-only access, Hermes gateway fix, Hermes claude setup, Hermes mcp setup, Hermes mem0 setup, Hermes railway setup, Hermes soul md setup, Hermes langfuse fix, and Hermes kanban setup.
WHAT I FIX AND BUILD:
- Owner-only fix (stop replying to everyone)
- Chat history and session repair
- Desktop install: Mac, Windows, Linux
- VPS deploy with auto-restart
- Telegram, Discord, WhatsApp gateways
- Claude Sonnet or Opus routing
- MCP: n8n, GitHub, Composio, Linear
- Mem0 memory, per-user isolation
- Railway plus Neon Postgres deploy
- SOUL.md, AGENTS.md, rules
- Langfuse tracing, cost tracking
- Kanban swarm, background subagents
TOOLS I USE:
Hermes Agent, Claude API, Docker, n8n, MCP, Mem0, Railway, Neon Postgres, Langfuse, Composio
Bot silent? Losing memory? Desktop stuck? Message me now, I will get it running today.
Get to know Anderson P
BUILD AUTOMATE SCALE AI Agent AI Engineer
- FromUnited States
- Member sinceMar 2026
- Avg. response time1 hour
Languages
English, French, Spanish
My Portfolio
FAQ
How do you fix Hermes replying to every group member instead of just me?
I set telegram_allowed_users to your numeric ID and correct BotFather privacy mode. This is a config fix, not a rebuild, and takes minutes once I confirm your gateway profile.
Can you set up Hermes with Mem0 so each team member has isolated memory?
Yes. I configure MEM0_USER_ID per teammate so memory never crosses between people, with a circuit breaker so a Mem0 outage does not take your agent down
Do you deploy Hermes on Railway with Neon Postgres for persistent memory?
Yes. I deploy Hermes with a Railway volume for config and SOUL.md, plus Neon Postgres over private networking so memory and sessions survive redeploys.
Can you configure Langfuse so I can see cost and traces per subagent?
Yes. I enable the bundled Langfuse plugin so every turn, tool call, and subagent delegation gets its own span with token cost, letting you audit spend.
Does the Hermes desktop app share memory and config with my VPS install?
No, they are separate installs unless you point both at the same profile. I can sync config, SOUL.md, and Mem0 memory so desktop and VPS share one identity and one memory store.
How do you structure SOUL.md and AGENTS.md so the agent stays on task?
I write a focused SOUL.md under 400 words for identity and rules, and push longer context into AGENTS.md, so the agent does not drift from its role.
Will you configure MCP so Hermes only uses native handlers, no gateway?
Yes. I can wire MCP servers directly through Hermes' native handler system without Composio or other aggregator layers, if that fits your security needs.
Can Hermes read and write Google Sheets or Excel during a conversation?
Yes. I connect a spreadsheet MCP tool so Hermes can pull data, log results, or update rows directly from a Telegram or Discord message.
Do you set up Kanban swarm mode for parallel multi-agent task execution?
Yes. I configure workers, a verifier, and a synthesizer so tasks run in parallel on a shared board instead of one agent doing everything in sequence.
Can you route Hermes through Claude Sonnet with a cost-aware fallback?
Yes. I set Claude as primary with a cheaper fallback provider, so Hermes stays fast on hard tasks but does not run up your bill on routine turns.

