Agent Almanac

What Is the Agent Almanac?

The Agent Almanac is a comprehensive, continuously updated directory of AI agents, MCP servers, agent tools, and agentic workflows published at almanac.metavert.io. Unlike conventional software directories curated exclusively by humans, the Agent Almanac is largely discovered, analyzed, and indexed by an AI agent itself—an autonomous system called the Agent(Agent)—making it one of the first agent-maintained registries of agents. Every entry passes through a review queue before publication, blending machine-speed discovery with human editorial oversight.

How Discovery Works

The Agent Almanac operates through multiple automated discovery channels. A Web Crawler monitors RSS feeds, GitHub repositories, and curated awesome-lists, streaming new content to Anthropic's Claude API for real-time analysis and classification. A dedicated Moltbook Agent browses the AI-only social platform Moltbook, searching for discussions about emerging tools and projects, maintaining persistent sessions, engaging with posts, and submitting discoveries back to the almanac. A HuggingFace Scanner watches trending Spaces to catch agents and tools gaining traction in the ML community. Additionally, anyone—human or AI—can submit entries through the web interface or a Machine API, and the system tags each entry with its discovery source for provenance tracking.

Why Agent Registries Matter

As the agentic economy scales—projected to exceed $50 billion by 2030—the ability to discover, evaluate, and compose agents becomes a critical infrastructure layer. A single brilliant agent is useful, but a network of composable agents that discover and delegate to each other is transformative. Agent registries like the Agent Almanac serve a function analogous to DNS for the internet: they let agents and humans locate the right capability at the right time. This concept has precedent in blockchain-based systems such as Fetch.ai's Almanac smart contract, which functions as a decentralized on-chain registry where autonomous agents register their addresses, endpoints, and capabilities. The emergence of open standards—including Google's Agent-to-Agent (A2A) Protocol, the IETF's AI Agent Discovery and Invocation Protocol (AIDIP), and the proposed Agent Name Service (ANS)—signals that agent discovery is maturing from proprietary silos into interoperable infrastructure. In February 2026, NIST launched the AI Agent Standards Initiative to ensure autonomous agents can interoperate securely across the digital ecosystem.

From Directory to Platform

The Agent Almanac reflects a broader shift in how software indexes are built and consumed. Traditional directories assume a human reader browsing categories; agent almanacs must also serve machine consumers that query programmatically, negotiate capabilities, and chain tools into agentic workflows without human intervention. By exposing a Machine API alongside its web interface, the Agent Almanac supports both paradigms—enabling a developer to browse agent listings in a browser while an orchestrator agent queries the same data to assemble a multi-agent pipeline. This dual-interface pattern is becoming the standard expectation for infrastructure in the agentic era, where the line between user and software agent increasingly blurs.

Relationship to the Agentic Economy

The Agent Almanac sits at the intersection of several forces reshaping technology markets. AI agents are making individual SaaS products subordinate to agentic workflows, collapsing the per-seat pricing model that built the SaaS industry. Venture capital investment in AI reached $211 billion in 2025—half of all global VC funding—with agentic AI representing the fastest-growing enterprise software segment. Within this landscape, discovery infrastructure like the Agent Almanac is not merely a convenience but a prerequisite: without reliable ways to find, verify, and compose agents, the agentic economy cannot achieve the network effects required for its projected scale. The Agent Almanac's model of agent-driven curation points toward a future where much of the knowledge work involved in maintaining directories, marketplaces, and indexes is itself delegated to autonomous agents.