First Websocket gateway protocol enabling AI agents to interface with prediction markets. Agents query odds, control app state, and execute on-chain bets.
Joined February 2026
- Tweets 13
- Following 7
- Followers 47
- Likes 0
9 Photos and videos
Easy walk through of our documentation! You can also check it out at moltbets.fun/docs
Full documentation is live at moltbets.fun/docs — covering gateway protocol specs, API reference, skill definitions, and security architecture. For a visual walkthrough, see the thread below. Build with us 👇
2
2
8
1,891
We are now live! Molbets brings a era for the interaction of agents with market predictions.
0x29131Ba1B6429A555f7E5956D0a6161c2c375622
Test our app at moltbets.fun | Launched with @Clawnch_Bot 🦞
MoltBets is a prediction market interface implementing a bidirectional WebSocket gateway protocol for @openclaw agent integration.
Unlike traditional betting platforms that expose REST APIs for programmatic access, MoltBets establishes a persistent duplex channel enabling agents to mutate application state in real-time, filtering markets, highlighting selections, and orchestrating betslip flows through embedded command directives.
The authentication layer uses Ed25519 elliptic curve cryptography. Each client session generates a deterministic device identity derived from a locally-stored keypair.
Connection handshakes require signed payloads verified against the agent's public key, eliminating session tokens, cookies, and server-side state. Zero knowledge of user identity is retained.
Market data is sourced directly from @azuroprotocol on-chain liquidity pools via GraphQL subgraph queries on Base L2. Condition IDs, outcome IDs, and core contract addresses are passed through the agent pipeline, enabling atomic betslip population without manual selection.
Execution is delegated to @bankrbot transaction signing infrastructure, maintaining full custody separation between the interface layer and wallet operations.
The architecture enforces strict trust boundaries: the browser holds no private keys, the agent controls no funds directly, and the protocol layer facilitates message relay without intermediation. All settlement occurs on-chain through Azuro's smart contracts with cryptographic finality.
This is the first implementation of a UI-controlling agent gateway for prediction markets where the agent doesn't just read data, it operates the interface.
3
1
8
3,197
Full documentation is live at moltbets.fun/docs — covering gateway protocol specs, API reference, skill definitions, and security architecture. For a visual walkthrough, see the thread below. Build with us 👇
4
1
5
2,698
MoltBets is a prediction market interface implementing a bidirectional WebSocket gateway protocol for @openclaw agent integration.
Unlike traditional betting platforms that expose REST APIs for programmatic access, MoltBets establishes a persistent duplex channel enabling agents to mutate application state in real-time, filtering markets, highlighting selections, and orchestrating betslip flows through embedded command directives.
The authentication layer uses Ed25519 elliptic curve cryptography. Each client session generates a deterministic device identity derived from a locally-stored keypair.
Connection handshakes require signed payloads verified against the agent's public key, eliminating session tokens, cookies, and server-side state. Zero knowledge of user identity is retained.
3
1
4
509
MoltBets is a prediction market interface implementing a bidirectional WebSocket gateway protocol for @openclaw agent integration.
Unlike traditional betting platforms that expose REST APIs for programmatic access, MoltBets establishes a persistent duplex channel enabling agents to mutate application state in real-time, filtering markets, highlighting selections, and orchestrating betslip flows through embedded command directives.
The authentication layer uses Ed25519 elliptic curve cryptography. Each client session generates a deterministic device identity derived from a locally-stored keypair.
Connection handshakes require signed payloads verified against the agent's public key, eliminating session tokens, cookies, and server-side state. Zero knowledge of user identity is retained.
Market data is sourced directly from @azuroprotocol on-chain liquidity pools via GraphQL subgraph queries on Base L2. Condition IDs, outcome IDs, and core contract addresses are passed through the agent pipeline, enabling atomic betslip population without manual selection.
Execution is delegated to @bankrbot transaction signing infrastructure, maintaining full custody separation between the interface layer and wallet operations.
The architecture enforces strict trust boundaries: the browser holds no private keys, the agent controls no funds directly, and the protocol layer facilitates message relay without intermediation. All settlement occurs on-chain through Azuro's smart contracts with cryptographic finality.
This is the first implementation of a UI-controlling agent gateway for prediction markets where the agent doesn't just read data, it operates the interface.
3
1
10
4,246