How to use Blablador with Hermes Agent¶
Helmholtz Blablador is not only the most useful and private Chat Robot out there. You can also connect it to Hermes Agent, to get a powerful CLI AI assistant with access to Blablador's private models. Imagine a full-featured AI agent that runs entirely in your terminal, with private model access!
Hermes Agent¶
Hermes is a CLI-based AI agent that can: - Execute terminal commands - Read and write files - Browse the web - Manage memory and skills - Delegate tasks to subagents - And much more...
All powered by Blablador's private models.
Prerequisites¶
- Hermes Agent installed
- A Blablador Token
Steps¶
1. Install Hermes¶
Follow the installation instructions on the Hermes Website. It's a one-liner copy-and-paste.
2. Configure Blablador Provider¶
When asked if you want to do a full configuration or a quick one, choose quick. The first option is the provider. For BLABLADOR, the choice is "Custom Endpoint (enter URL manually)". Then enter the base URL:
It will ask for your API key. Paste your Blablador token here. The token will NOT appear on the screen for security reasons, but it is being saved.
The next question is the model alias. You can choose any of the Blablador aliases (alias-code, alias-fast, alias-large, alias-huge). For most use cases, alias-code is a good choice.
Next is Display Name. You can choose "Blablador" or any name you like.
Soon after, it will ask if you want to start chat. You can!
3. Setting up a search engine (optional, but VERY useful)¶
As soon as you start chatting, you can paste the following instructions:
Add a new skill for web search. It will use the searxng search engine with the following URL: https://search.blablador.fz-juelich.de - and this will be your MAIN AND ONLY web search. You should name it blablador-search- Test it afterwards.
Adding a CLAUDE.md-like behavioral guideline¶
Some models benefit from having a behavioral guideline to reduce common mistakes. You can create a file called $HOME/.hermes/CLAUDE.md with the following content:
# CLAUDE.md
Behavioral guidelines to reduce common LLM coding mistakes. Merge with project-specific instructions as needed.
**Tradeoff:** These guidelines bias toward caution over speed. For trivial tasks, use judgment.
## 1. Think Before Coding
**Don't assume. Don't hide confusion. Surface tradeoffs.**
Before implementing:
- State your assumptions explicitly. If uncertain, ask.
- If multiple interpretations exist, present them - don't pick silently.
- If a simpler approach exists, say so. Push back when warranted.
- If something is unclear, stop. Name what's confusing. Ask.
## 2. Simplicity First
**Minimum code that solves the problem. Nothing speculative.**
- No features beyond what was asked.
- No abstractions for single-use code.
- No "flexibility" or "configurability" that wasn't requested.
- No error handling for impossible scenarios.
- If you write 200 lines and it could be 50, rewrite it.
Ask yourself: "Would a senior engineer say this is overcomplicated?" If yes, simplify.
## 3. Surgical Changes
**Touch only what you must. Clean up only your own mess.**
When editing existing code:
- Don't "improve" adjacent code, comments, or formatting.
- Don't refactor things that aren't broken.
- Match existing style, even if you'd do it differently.
- If you notice unrelated dead code, mention it - don't delete it.
When your changes create orphans:
- Remove imports/variables/functions that YOUR changes made unused.
- Don't remove pre-existing dead code unless asked.
The test: Every changed line should trace directly to the user's request.
## 4. Goal-Driven Execution
**Define success criteria. Loop until verified.**
Transform tasks into verifiable goals:
- "Add validation" → "Write tests for invalid inputs, then make them pass"
- "Fix the bug" → "Write a test that reproduces it, then make it pass"
- "Refactor X" → "Ensure tests pass before and after"
For multi-step tasks, state a brief plan:
1. [Step] → verify: [check]
2. [Step] → verify: [check]
3. [Step] → verify: [check]
Strong success criteria let you loop independently. Weak criteria ("make it work") require constant clarification.
---
**These guidelines are working if:** fewer unnecessary changes in diffs, fewer rewrites due to overcomplication, and clarifying questions come before implementation rather than after mistakes.
Comparison with OpenCode¶
While OpenCode provides a Claude Code-like experience with open-source models, Hermes offers:
- Full CLI agent with extensive toolset
- Memory system for long-term context
- Skills framework for reusable workflows
- Subagent delegation for parallel tasks
- Cron jobs for automated tasks
- Multi-platform support (CLI, Telegram, Discord, etc.)
Choose Hermes when you need a full-featured AI assistant that can execute complex, multi-step tasks autonomously.
References¶
And that's it! You now have a powerful, private AI agent running in your terminal with access to Blablador's models. Enjoy!