Hermes Agent

Connect Hermes Agent to Requesty with the native Requesty plugin for access to 600+ models, automatic prompt caching, cost tracking, and fallback routing.

Hermes Agent is an open source AI agent by Nous Research that runs across CLI, TUI, desktop, and messaging platforms. It features persistent memory, tool use, browser automation, and multi-session delegation.

Requesty ships as a native provider in the Hermes plugin catalog, so you install it with one command and pick Requesty in the same model picker as every other provider. No YAML to write.

Using the Requesty integration, you can:

• Access 600+ models from OpenAI, Anthropic, Google, Mistral, and many other providers through one API key.
• Get automatic prompt caching on long multi-turn sessions with nothing to configure.
• Track and manage your spend in a single location, grouped per Hermes session.
• Apply fallback policies, load balancing, and latency routing to keep your agent responsive.

Prerequisites

• Hermes Agent version 0.21.6 or newer (curl -fsSL https://hermes-agent.nousresearch.com/install.sh | bash). Check with hermes --version.
• A Requesty API key from the API Keys page.

Setup

Pick the surface you use. The plugin is the same in all three, only the clicks differ.

Terminal (CLI)

1. Install the Requesty plugin

   hermes plugins install requesty --enable
   

   Hermes downloads the plugin from its catalog and prints a short after-install guide. If you installed without --enable, run hermes plugins enable requesty.

2. Add your API key

   Hermes keeps secrets in ~/.hermes/.env:

   echo 'REQUESTY_API_KEY=your_requesty_api_key' >> ~/.hermes/.env
   
3. Pick Requesty as your provider

   hermes model
   

   Choose Requesty in the provider list, then a model. The list is Requesty’s catalog of managed models that support tool calling, fetched live from your account.

   Or start a session with the provider and model on the command line:

   hermes --provider requesty -m claude-sonnet-5-5
   
4. Verify

   hermes chat -q "Which model and provider are you running on?"
   

   Then open the Requesty analytics dashboard. The request appears with Hermes as its origin. hermes doctor also lists Requesty under API Connectivity.

TUI

1. Install the plugin and add your key

   Run once in a terminal:

   hermes plugins install requesty --enable
   echo 'REQUESTY_API_KEY=your_requesty_api_key' >> ~/.hermes/.env
   
2. Open the model picker

   Start the TUI with hermes --tui, type /model and press Enter. Type req to filter the provider list and select Requesty.

   
3. Pick a model

   The second step lists Requesty’s managed models. Choose one and press Enter.

   
4. Chat

   Your next message goes through Requesty. The status bar shows the model, context usage and the estimated cost per turn.

   

Desktop app

No terminal needed. Everything is done in the app.

1. Find Requesty in the plugin catalog

   Open Capabilities in the left sidebar, then the Plugins tab. In the plugin catalog at the bottom, search requesty and click + Add to this Agent.

   
2. Install

   Confirm with Install. The plugin is reviewed by Nous and pinned to a verified version. Hermes installs it and switches it on for the current profile.

   
3. Add your API key

   Open Settings → Providers → API keys, expand the Requesty row, paste your key and press Save. Get a key opens the Requesty API keys page. The optional Requesty base URL override takes https://router.eu.requesty.ai/v1 if you want all traffic on the EU router.

   
4. Pick a model

   In the chat composer, click the model name at the bottom right. Requesty’s models are listed under REQUESTY.

   
5. Chat

   Send a message. It goes through Requesty and shows up in your analytics dashboard with Hermes as its origin.

   

The gateway (Telegram, Discord, Slack and the other platforms) uses whatever default you set with hermes model.

Model selection

The picker shows Requesty’s managed model IDs, for example claude-sonnet-5-5, gpt-6.1-sol, gemini-3.8-flash, kimi-k3. These resolve on the Requesty side to the best available provider for that model, with fallback built in.

You can also type any other ID from the Model Library directly:

hermes --provider requesty -m anthropic/claude-sonnet-4-5
hermes --provider requesty -m policy/reliable-sonnet

Hermes reasoning levels (/reasoning low|medium|high) are sent to Requesty as reasoning_effort on models that support it.

Recommended: use a Routing Policy

Instead of hard coding a model, point Hermes at a Routing Policy. A policy is a named alias that resolves on the Requesty side. You swap the underlying model from the Routing Policies page without touching Hermes.

hermes --provider requesty -m policy/reliable-sonnet

Policy types that work well with Hermes:

• Fallback Policy for reliability. If your primary model is down, Requesty retries the next in the chain.
• Latency Routing for speed. Requesty picks whichever provider is currently fastest.
• Load Balancing for gradual rollouts between models.

Prompt caching

Hermes sends a large system prompt (often 15,000+ tokens including tool definitions) on every turn. Requesty recognises Hermes traffic and applies automatic caching server side, so from the second turn onwards the system prompt and tool definitions are served from cache. There is nothing to configure on the Hermes side. You can see the effect as cached_tokens in the analytics dashboard and in Hermes Desktop’s cache hit rate indicator.

Cost and tracing

• Each Hermes session ID is sent as requesty.trace_id, so one conversation groups into a single trace in the Requesty logs.
• Hermes shows an estimated cost per turn using Requesty’s public model catalog. The billed amount is in your Requesty dashboard.

EU routing

To keep all traffic in the EU region, point the plugin at the EU endpoint in ~/.hermes/.env:

REQUESTY_BASE_URL=https://router.eu.requesty.ai/v1

Alternatively, pick an @eu model variant such as claude-sonnet-5-5@eu in the model picker to route only that model through the EU. See EU Routing for details.

Manual configuration (Hermes older than 0.21.6)

Older Hermes versions do not support provider plugins. You can still connect through a custom_providers entry in ~/.hermes/config.yaml:

model:
  default: "anthropic/claude-sonnet-4-5"
  provider: "requesty"

custom_providers:
  - name: requesty
    base_url: "https://router.requesty.ai"
    api_key: "your_requesty_api_key"
    api_mode: anthropic_messages

The Requesty CLI writes this manual configuration for you: run requesty, select Hermes, pick a model, and it backs up your existing config before writing.

Troubleshooting

Requesty does not appear in hermes model

Run hermes plugins list and confirm requesty is installed and enabled. If it is disabled, run hermes plugins enable requesty. The plugin requires Hermes 0.21.6 or newer; run hermes update if you are on an older build.

401 or 403 Invalid authorization token

Verify your API key is correct and active on the API Keys page, and that REQUESTY_API_KEY in ~/.hermes/.env matches exactly, including any trailing = characters. Restart Hermes after editing .env.

Model not found

Check that the model ID is available in the Model Library. If your organization uses approved models, ensure the model is on the approved list. The picker only lists managed models that support tool calling; other IDs can still be passed with -m.

Caching not working (cached_tokens stays at 0)

Caching starts from the second turn of a conversation, so a single hermes chat -q call will not show cached tokens. Make sure the model you picked supports prompt caching (see the Model Library).

References

• Hermes Agent docs
• Hermes plugin catalog
• Requesty plugin for Hermes on GitHub
• Requesty API Keys
• Requesty Model Library
• Requesty Auto-Caching