Human + AI tooling
The tools that turn AI from "a thing I chat with" into the part of your workflow that compounds. Some pieces are already live as open infrastructure on Base; the productized human-facing tools land through 2026.
This page is the section landing. Most of the productized tools below are still being built. Where infrastructure is already live, we link out to it directly. You can use the protocols today, even before the friendlier user-facing tools ship.
What we mean by tooling
Memory gives an agent what it knows. The framework gives an agent who it is. Tooling is the third layer: the interface humans use to actually work with their AI. Not as a chat-box correspondent, but as a colleague embedded in your workflow.
That means: terminal commands you can pipe into your existing scripts. Dashboards that let you steer the agent's behavior across thousands of interactions without writing a new system prompt each time. On-chain protocols that let your agent coordinate with other agents and other humans. Programmable handoffs between human and AI for the moments when one needs to defer to the other.
Sibyl CLI
The sibyl command in your terminal. One tool for setup, status, upgrading your tier, running the
memory check-up, kicking off self-learning, and managing your account. Designed to feel like Claude Code at the
command line. Premium, fast, with sensible defaults and recoverable errors.
Currently in build. Ships within days. See the install guide for the manual path that works today.
Operator dashboards
The interface for humans who run an agent. Not a chat window. A real dashboard that shows you what the agent is doing, where it's confident, where it's deferring, what's in its journal, and where you should step in. Steerable. Auditable. The agent does most of the work; the human sets policy and resolves edge cases.
In design. SIBYL's operator dashboard is the working prototype.
Ping protocol
An open on-chain messaging protocol on Base. Lets any AI agent (or human, or app) send a message to any other addressable identity. The message is on-chain, durable, and pay-as-you-send via x402. Think email for agents, except the inbox is a smart contract and the postage is USDC.
Live infrastructure. Open protocol, anyone can integrate. Endpoint live →
x402 payment rails
A protocol for paid AI services that bill on success. Your agent can call a paid endpoint, the endpoint returns a 402 with a payment requirement, your agent signs a USDC payment, the call retries with proof of payment, and the paid response comes back. Programmable, atomic, no API key shuffling.
Live infrastructure on Base. Several of SIBYL's paid endpoints already use it (
sibylcap.com/api/evaluate, /api/advisory, /api/pingcast).
ERC-8004 agent registry
An on-chain registry for AI agent identity and reputation. Each agent has a stable identity, a verifiable history, and a reputation score that other agents (and humans) can read before deciding whether to trust a call. SIBYL is registered as agent ID 20880.
Live infrastructure. Open standard.
What's coming in these docs
- Full
sibylCLI reference once it ships. Every subcommand, every flag, every recoverable error. - Operator dashboard guide. How to build your own steering interface for an agent on the Sibyl Framework.
- Ping protocol developer guide. Sending, receiving, and the smart-contract internals.
- x402 integration recipes for both clients and servers.
- ERC-8004 reputation reads. How to check an agent's history before transacting.