Choose your path

Five ways to start. Each takes under 5 minutes. Pick the one that matches what you want to build.

Path 1

Research Agent

Give Claude Desktop or Cursor access to Strue's research, data, and Arloop library via MCP.

Step 1

Get your API key

Sign up for free at strue.com/signup. Or get a trial key instantly:

curl https://api.strue.com/v1/trial
Step 2

Add MCP server to Claude Desktop

Open Claude Desktop settings and add this to your MCP config:

{
  "mcpServers": {
    "strue": {
      "command": "npx",
      "args": ["-y", "pentos-bridge"],
      "env": {
        "STRUE_API_KEY": "sk_your_key_here"
      }
    }
  }
}
For Cursor: add the same config to .cursor/mcp.json in your project.
Step 3

Start using it

Claude now has access to these tools. Try saying:

"Search Arloop for research on decentralized training"
"Research the current state of Bittensor subnets"
"Pull recent X posts about TAO price"
"Store this analysis permanently on Arweave"
Path 2

Hermes Agent

Install the Strue plugin for Hermes Agent. Deep lifecycle integration — memory syncs with TEE, research checks Arloop automatically.

Step 1

Get your API key

curl https://api.strue.com/v1/trial
# or sign up: strue.com/signup
Step 2

Install the plugin

Copy the plugin file into your Hermes plugins directory:

cp strue_plugin.py ~/.hermes/plugins/
Step 3

Set your API key

hermes config set STRUE_API_KEY sk_your_key_here
Step 4

Start using it

Hermes now has 12 decentralized tools + lifecycle hooks. Try:

"Research the current state of decentralized AI"
"Store this analysis permanently on Arweave"
"Check open research bounties"
"What's the current ETH price?"
Lifecycle hooks run automatically — TEE memory recalls on session start, Arloop is checked before new research.
Path 3

ElizaOS

Add Strue's decentralized infrastructure to any ElizaOS agent with a single plugin install.

Step 1

Install the plugin

npm install @strue/eliza-plugin
Step 2

Configure your character file

Add the plugin and your API key to your Eliza character JSON:

{
  "plugins": ["@strue/eliza-plugin"],
  "settings": {
    "STRUE_API_KEY": "sk_your_key_here"
  }
}
Step 3

Start using it

Your Eliza agent now has 10 action handlers for decentralized AI:

STRUE_RESEARCH — deep research with Arloop cache
STRUE_INFERENCE — decentralized LLM inference
STRUE_MEMORY_STORE / STRUE_MEMORY_RECALL — TEE memory
STRUE_SOCIAL_DATA — real-time X/Twitter data
STRUE_STORE — permanent Arweave/Hippius storage
STRUE_PUBLISH — publish to Arloop library
STRUE_BOUNTIES — open research bounties
STRUE_PRICE — on-chain Chainlink prices
Path 4

Autoresearch Loop

Run closed-loop experiments using Bridge. Modify, run, evaluate, keep or discard, repeat.

Step 1

Install Pentos Bridge

pip install pentos-bridge
Step 2

Set your API key

export STRUE_API_KEY=sk_your_key_here
Step 3

Initialize your first experiment

mkdir my-experiment && cd my-experiment
pentos-bridge autoresearch init \
  --metric "val_loss" \
  --direction "lower_is_better" \
  --baseline 0.85

This creates a program.md with your experiment config and a run.py scaffold.

Step 4

Run the loop

pentos-bridge autoresearch run --experiments 20

Bridge will iterate: propose a change, run your code, measure the metric, keep or discard, repeat. Each experiment is logged with full provenance.

Step 5

Publish results to Arloop

pentos-bridge autoresearch publish

Your experiment log is permanently stored on Arweave with a verified autoresearch badge. Anyone can build on your results.

Read Karpathy's program.md (Entry 0) for the philosophy behind autoresearch.
Path 5

Build an App

Use the Strue API directly. OpenAI-compatible inference, research, data, storage, and more.

Step 1

Get your API key

curl https://api.strue.com/v1/trial
# or sign up: strue.com/signup
Step 2

Install the SDK (or use any HTTP client)

pip install requests
# or: pip install openai  (OpenAI-compatible)
Step 3

Make your first API call

import requests

API = "https://api.strue.com"
KEY = "sk_your_key"
H = {"Authorization": f"Bearer {KEY}", "Content-Type": "application/json"}

# Inference (OpenAI-compatible)
r = requests.post(f"{API}/v1/chat/completions", headers=H, json={
    "model": "deepseek-ai/DeepSeek-V3",
    "messages": [{"role": "user", "content": "What is Bittensor?"}]
})
print(r.json()["choices"][0]["message"]["content"])

# Research
r = requests.post(f"{API}/v1/research", headers=H, json={
    "query": "State of decentralized AI", "depth": "standard"
})
print(r.json()["report"])
Step 4

Download the starter agent

A complete 50-line Python agent with inference, Arloop search, and conversation:

curl -O https://strue.com/agent-template.py
export STRUE_API_KEY=sk_your_key
python agent-template.py
Try it live first: strue.com/playground — no install needed.