Bootstrap Hermes with PayPerQ (ppq.ai)

1 · Install Hermes

curl -fsSL https://hermes-agent.nousresearch.com/install.sh | bash
hermes --version

On the first hermes launch the TUI runs you through basics (name, theme, telemetry).

2 · Install the ppq plugin (from its GitHub repo)

The plugin is not yet in the Hermes plugin catalog, so install it manually from the repository. This copies the plugin files into Hermes' plugin directory — exactly where they live on the author's machine (verified identical to the repo):

git clone https://github.com/thepurpose-npub/ppq-hermes-provider
mkdir -p ~/.hermes/plugins/model-providers/ppq
cp ppq-hermes-provider/model-providers/ppq/{__init__.py,plugin.yaml} \
   ~/.hermes/plugins/model-providers/ppq/

Optional, pinned install: cd ppq-hermes-provider && git checkout ca8cea59a3712cffa31d2759a30a9f8be66fddc2 before the cp — installs exactly the version the author runs.

3 · Get a PayPerQ account + API key

Pay-per-use, no subscription: top up from $0.10 (crypto or card), then grab a key.

• Fastest, no email: curl -X POST https://api.ppq.ai/accounts/create → returns a credit_id + api_key immediately. Top up that credit_id at ppq.ai.
• Or sign up in the browser at ppq.ai and create the key under API Docs.

Your key = sk-…. Keep it private.

4 · Store the API key in Hermes' env file

The plugin reads PPQ_API_KEY from the environment (its documented auth path):

hermes config env-path      # prints the env file location
# add this line to that file (create it if missing):
# PPQ_API_KEY=sk-your-key-here

Then restart Hermes (quit the TUI, start hermes again — or hermes gateway restart if you run a gateway) so the plugin loads and picks up the key.

Verify it loaded:

hermes plugins list         # should show "ppq — PayPerQ — v0.3.0"

Alternative (optional): hermes auth add ppq stores the key in Hermes' credential pool instead of the env file. Either works — the env file is the plugin README's documented path.

5 · Pin glm-5.3-flash as your only pinned model

The plugin's picker resolves pins in layers: your pins.json wins over any live/static defaults, so write exactly one entry.

mkdir -p ~/.hermes/ppq-plugin
cat > ~/.hermes/ppq-plugin/pins.json <<'EOF'
[
  "z-ai/glm-5.3-flash"
]
EOF

Then launch hermes, run /model and select the ppq provider — z-ai/glm-5.3-flash should be the only entry under its pinned section. Select it so it becomes the default.

Per the author's instruction, this is the only pinned model — glm-5.3-flash is not a trial/free model, so no bootstrap-credit fallback is configured here. If you'd rather browse the live catalog first, delete pins.json and the plugin falls back to a curated list of top chat models.

6 · Smoke test

hermes "Say hello in one sentence."

Should return a reply priced at roughly $0.002–$0.01 from your credit balance (glm-5.3-flash is $0.16/M input, $0.53/M output — very cheap). If the cost readout shows an exact number, the plugin's pricing hook is working.

7 · LLM handover — from here on, talk to Hermes

Once the smoke test works, the agent itself is the LLM interface. From here on you can:

• Just type questions/commands — Hermes plans and executes using its built-in tools (web search, files, terminal, etc.).
• Hand over any concrete task in plain language — e.g. "summarize this URL", "research X and give me 3 options", "write a script that does Y".
• Use /new when switching topics, to keep sessions clean and cheap.
• /model to switch models later (the plugin also exposes a live list from ppq's /v1/models).

Troubleshooting