Skip to main content
Back

MiroShark Simulates 100 AI Agents Reacting to Your Content for $1

MiroShark spawns hundreds of AI personas to simulate public reaction to any document — press releases, policy drafts, or financial news — for under $1.

Repo, Full Access
  • What it is: MiroShark is an open-source engine that spawns hundreds of AI personas to simulate public reaction to any document.
  • Cost and speed: A full simulation costs ~$0.30 on Qwen3 235B and returns results in under 10 minutes.
  • How it works: Extracts a Neo4j knowledge graph from your document, generates agents with unique biases and influence levels, then runs them across simulated Twitter, Reddit, and a prediction market.
  • Powered by OASIS: The simulation engine is CAMEL-AI's OASIS, a research-grade simulator capable of modeling up to one million agents.
  • Flexible deployment: Runs locally via Ollama, via cloud APIs (OpenAI, Anthropic, OpenRouter), or as a hosted $1/run x402 endpoint at miroshark.xyz.
  • Key limitation: Synthetic agents cannot replicate lived human experience; outputs should be used for stress-testing assumptions, not as ground truth predictions.

MiroShark is an open-source swarm intelligence engine that does something previously reserved for expensive PR firms: it simulates how the internet would react to your content, before you publish it. Feed it a press release, a policy draft, a financial report, or even a tweet, and it spins up hundreds of AI agents that argue, post, shift opinions, and go viral , hour by hour , on simulated versions of Twitter, Reddit, and a prediction market.

MiroShark workflow: input document to swarm of 100 agents to output report, for $1 in under 10 minutes

The $1 focus group that never existed

MiroShark lets you stress-test narratives before they hit the real world. Upload a press release, policy draft, or financial report and watch hundreds of AI-driven personas simulate social media reactions, debates, and opinion shifts hour by hour. The whole thing runs in under 10 minutes and costs roughly $0.30 on Qwen3 235B via OpenRouter , well under the $1 ceiling the project advertises.

The project is an English fork of MiroFish, originally built by 666ghj at Shanda Group. Powered by the OASIS simulation engine from CAMEL-AI, it runs locally with Ollama or via cloud APIs. It features improved simulation flows, recommended models, and full local deployment options under the GNU AGPL v3 license.

How it actually works

The pipeline has five distinct phases, each doing meaningful work before passing state to the next:

  1. Ontology Generation , The engine reads your document and extracts a schema of all relevant entities and relationships.
  2. Graph Building , MiroShark builds a Neo4j knowledge graph that captures entities, relationships, and per-agent memory.
  3. Agent Setup , Hundreds of AI agents are generated with unique personalities, biases, and influence levels. Each agent gets a demographic seed, web-enriched background, and semantic relationships to other agents in the graph.
  4. Simulation Execution , Agents interact across simulated environments representing Twitter, Reddit, and Polymarket, tracking belief states, confidence levels, and trust metrics over time. All three platforms run simultaneously. A single LLM-generated prediction market with non-50/50 starting prices drives Polymarket trading. Agents see cross-platform context: traders read Twitter/Reddit posts, social media agents see market prices.