Skip to content

084-farcaster-ai-agents-landscape

Directory actions

More options

Directory actions

More options

Latest commit

 

History

History

084-farcaster-ai-agents-landscape

Folders and files

NameName
Last commit message
Last commit date

parent directory

..
README.md
README.md
 
 

README.md

84 — Farcaster AI Agents Landscape (March 2026)

topic: agents type: research status: research-complete last-validated: 2026-05-21 original-query: Map the landscape of AI agents and bots on Farcaster for building ZAO community bot (reconstructed) tier: reference

Status: Research complete Date: March 19, 2026

Context

Deep research on the full landscape of AI agents, bots, and automated accounts on Farcaster as of March 2026. Conducted for building a ZAO community bot.

1. Major Structural Change: Neynar Acquires Farcaster (January 2026)

On January 21, 2026, Neynar acquired Farcaster from Merkle. Founders Dan Romero and Varun Srinivasan stepped back. Neynar now maintains the protocol, runs the Farcaster client app, and operates Clanker.

What this means for bot builders:

  • Neynar is now both the protocol maintainer AND the primary API provider -- a single point of contact
  • A new builder-focused roadmap is expected
  • No immediate product changes to Farcaster, Neynar APIs, or Clanker
  • Centralization concerns exist: user data remains onchain, but application control sits with one company
  • Farcaster generated only $1.84M in Q4 2025 revenue (down 85% YoY), motivating this shift

Sources:

2. Catalog of Known AI Agents & Bots on Farcaster

Tier 1: Major Agents

Clanker (Token Bot)

  • What: AI-powered autonomous agent for no-code ERC-20 token deployment on Base
  • How: Users tag @clanker in a Farcaster cast, describe their token idea in natural language. Clanker deploys the token, creates a Uniswap V3 liquidity pool, locks LP forever
  • Scale: Over $7.62B all-time trading volume, 21,870 tokens created on a single day (Feb 2, 2026), $364M daily trading volume ATH
  • Revenue: 1% fee on trades; 80% of LP fees go to token creators; protocol uses remainder for CLANKER buybacks
  • Tech: Audited smart contracts, Clanker v4 introduced modular contract architecture and dynamic fee mechanisms
  • Farcaster acquired Clanker in Oct 2025 (now under Neynar)
  • Sources: https://coinmarketcap.com/cmc-ai/tokenbot-2/what-is/, https://clanker.gitbook.io/clanker-documentation/general/token-deployments/farcaster-bot-deployments

Aether

  • What: AI community manager + treasurer for onchain communities (created by Martin for HIGHER community)
  • How: Creates and sells NFTs, issues bounties on Bountycaster, manages treasury, deploys tokens via Clanker, forms social graphs
  • Scale (May 2026): NFT minted ~466K times, 6.8K+ token holders, treasury >$150K from NFTs + donations + subscriptions
  • Notable: Issued 59+ bounties on Bountycaster, created Luminous ($LUM) token, agents can now spend + earn + trade via X402 protocol (machine-to-machine transactions)
  • Impact: Demonstrates agents as economic stakeholders — Aether's treasury grows through autonomous decision-making
  • Source: https://www.bankless.com/farcaster-ai-agents-hub (April 2026); Medium articles on agent economics and decentralized social

AIXBT

  • What: Trading alpha / sentiment analysis agent
  • How: Scans social graphs on Farcaster, X, and Lens to spot early trends; provides market insights
  • Scale: Market value approached $100M
  • Source: https://www.aicoin.com/en/article/431385

Bankr

A0x (Onchain Minds)

Tier 2: Utility Agents

Gina (@askgina.eth)

Atlas

Bracky

@indexer

@degenbot

@remindme

Neynar AI

Bountycaster

TrueCastAgent

Farcaster Support Agent

Caster Agents Framework V3

Tier 3: Music-Adjacent (Relevant to ZAO)

Sonata

Sound.xyz Integration

No dedicated music AI agent exists on Farcaster as of March 2026. This is a clear opportunity for ZAO.

3. Farcaster Bot Policy & Rules

There is no official centralized bot policy. Farcaster uses a decentralized moderation approach:

What's Allowed

  • A Farcaster bot is just a regular Farcaster account -- no special status or restrictions
  • Automated replies based on @mentions are standard and accepted
  • Bots can post, reply, like, recast, follow, and interact with channels
  • Agents are explicitly supported by the ecosystem (Neynar has an "Agents and bots" section in their dev portal)

Anti-Spam Mechanisms

  1. Storage Rent: Accounts must pay storage rent (~$0.30/unit as of July 2025, previously $7/unit). Each unit lasts 1 year. This serves as an economic anti-spam barrier.
  2. Cast Volume Limits: Accounts creating >1,000 casts/24h need 20% of past 24h cast volume in available storage
  3. Frame-Based Filtering: Apps can filter bot interactions using onchain criteria (e.g., NFT holdings)
  4. Channel Moderation: Channel owners can set 20+ composable moderation rules, including anti-sybil controls
  5. Community Self-Policing: Open-source design lets users build their own bot-detection/exclusion tools

Known Issues

  • Bot activity reportedly inflates DAU metrics (reported 40K-60K DAUs, but only ~4,360 Power Badge holders)
  • Some users report "shadow-banning" concerns
  • Vitalik Buterin publicly commented on increased spam levels
  • ModBot (https://modbot.sh/) exists as a channel moderation helper

Sources:

4. How Neynar Supports Bot Development

Signer Options for Bots

Neynar offers three signer approaches:

Signer Type Best For Cost Notes
Sign In with Neynar (SIWN) Most apps, user-facing Free Pre-onboarded 35K+ users, plug-and-play
Neynar-Managed Signers Bots, agents, custom branding Free (no gas) Neynar stores keys, you get full API access. Recommended for bots.
Developer-Managed Signers Maximum control Highest burden You store keys and sign messages yourself

For a ZAO bot: Use Neynar-Managed Signers. Navigate to dev portal > "Agents and bots" > "use existing account" > connect bot wallet > receive signer UUID.

Bot Setup Flow

  1. Create a Farcaster account for the bot
  2. Get a Neynar API key
  3. Register a managed signer via the Neynar dev portal
  4. Use the signer UUID + API key to publish casts via the SDK
  5. Set up webhooks for mention detection and reply triggers

Webhooks

  • Trigger on specific Farcaster events (mentions, replies, follows)
  • Available on Professional plan ($49/mo) and above
  • Neynar AI's own agent uses this: webhook fires when @neynar is tagged, calls an API, generates response

API Pricing (March 2026)

Plan Price Compute Units Webhooks Key Limits
Free $0 200K No Basic read/write
Starter $9/mo 1M No Read/write, 300 RPM
Professional $49/mo 10M Yes 600 RPM, indexer-as-a-service
Enterprise $249/mo 60M Yes 1200 RPM, SQL playground, custom pipelines
Custom Contact Custom Yes Custom

Rate Limits

Plan Per-Endpoint Global (all APIs)
Starter 300 RPM / 5 RPS 500 RPM
Growth 600 RPM / 10 RPS 1000 RPM
Scale 1200 RPM / 20 RPS 2000 RPM
  • /validate, /signer, and signer/developer_managed APIs are exempt from global limits
  • Cast search is more restricted: 60-240 RPM depending on plan
  • Frame validation: 5K-20K RPM depending on plan
  • Rate limits are request-based, independent of credit/compute unit usage

Cast Volume Limits (Protocol-Level)

  • <1,000 casts/24h: No restrictions

  • 1,000 casts/24h: Must have 20% of past 24h volume in available storage

Source: https://docs.neynar.com/reference/what-are-the-rate-limits-on-neynar-apis, https://docs.neynar.com/docs/which-signer-should-you-use-and-why

5. Frameworks for Building Farcaster Bots

ElizaOS (Recommended for ZAO)

Current state: ElizaOS + Virtuals Protocol are the standard frameworks for building autonomous AI agents as of early 2026. ElizaOS has 90+ plugins, managing over $25M in AUM across deployed bots, with ecosystem partners at $20B+ combined market cap.

Farcaster Plugin (@elizaos/plugin-farcaster)

Features:

  • Autonomous publishing with configurable intervals (POST_INTERVAL_MIN/MAX, default 90-180 min)
  • Mention monitoring and auto-reply (FARCASTER_POLL_INTERVAL, default 2 min)
  • Like, recast, follow automation
  • Dry-run mode for testing without live posting
  • Built-in caching for profiles and cast data
  • Context-aware responses with memory/conversation tracking
  • Channel-specific posting
  • Cross-platform coordination (Discord, Telegram, X, Farcaster simultaneously)

Required env vars:

FARCASTER_NEYNAR_API_KEY=your-neynar-api-key
FARCASTER_SIGNER_UUID=your-signer-uuid  (from Neynar managed signer)
FARCASTER_FID=12345

Optional env vars:

FARCASTER_DRY_RUN=false          # Test mode
MAX_CAST_LENGTH=320              # Character limit
FARCASTER_POLL_INTERVAL=2        # Mention check frequency (minutes)
ENABLE_POST=true                 # Toggle autonomous posting
POST_INTERVAL_MIN=90             # Min minutes between posts
POST_INTERVAL_MAX=180            # Max minutes between posts
ENABLE_ACTION_PROCESSING=false   # Like/recast automation
ACTION_INTERVAL=5                # Action processing frequency (minutes)
POST_IMMEDIATELY=false           # Post on startup
MAX_ACTIONS_PROCESSING=1         # Interactions per cycle

Architecture:

  1. FarcasterClient -- Core Neynar API communication
  2. FarcasterPostManager -- Automated scheduling and cast generation
  3. FarcasterInteractionManager -- Mention and reply orchestration
  4. Memory Systems -- Conversation context storage

Character config:

import { Character } from "@elizaos/core";
import { farcasterPlugin } from "@elizaos/plugin-farcaster";

export const character: Character = {
  name: "ZAOBot",
  plugins: [farcasterPlugin],
  bio: "The ZAO community music agent",
  description: "I help discover and share music in the ZAO community"
};

Custom templates: farcasterPostTemplate, farcasterMessageHandlerTemplate, farcasterShouldRespondTemplate

Error handling: Implement retry with exponential backoff for rate limits. Validate cast length (320 char max).

Note: The old client-farcaster package is archived. Use @elizaos/plugin-farcaster (renamed with plugin- prefix).

Sources:

Neynar SDK (Direct)

Simpler approach for custom bots without ElizaOS:

Caster Agents Framework V3

Botcaster

Custom (Direct Hub + Neynar)

6. Farcaster Mini Apps + Agent Integration

Mini Apps (formerly Frames v2) are HTML/CSS/JS apps that run inside the Farcaster feed.

Agent-Relevant Capabilities

  • SDK: @farcaster/miniapp-sdk provides Ethereum Provider (EIP-1193), Quick Auth (Sign in with Farcaster + JWT), notifications
  • Agent Checklist: Official documentation at https://miniapps.farcaster.xyz/docs/guides/agents-checklist
  • Manifest: Must be at /.well-known/farcaster.json with fc:miniapp meta tags
  • Ready Signal: Apps must call sdk.actions.ready() after initialization
  • Embed Metadata: Version "1", imageUrl (3:2 ratio), button title max 32 chars

Key Rules for AI Agents Building Mini Apps

  • NEVER reference Frames v1 syntax (fc:frame:image, fc:frame:button)
  • Use fc:miniapp meta tags, not fc:frame
  • Always verify fields against official SDK schema
  • Use miniapp or frame (not frames) in manifests

Examples

  • TrueCastAgent: Powers prediction markets mini app
  • Neynar AI: In-feed mini app agent

Source: https://miniapps.farcaster.xyz/docs/guides/agents-checklist

7. Cost Structure for Bot Accounts

One-Time Costs

  • FID Registration: Requires renting 1 storage unit (~$0.30 in ETH as of July 2025)
  • Storage Rent: ~$0.30/unit/year (was $7/unit before July 2025 repricing)

Ongoing Costs

  • Neynar API: $0 (Free, 200K compute units) to $249/mo (Enterprise, 60M compute units)
  • For a community bot: Professional plan at $49/mo is the sweet spot (webhooks required for mention-triggered replies)
  • LLM costs: Depends on model (GPT-4o, Claude, etc.) -- typically $5-50/mo for a community bot
  • Hosting: Standard server costs for the bot process (Vercel, Railway, etc.)

Total Estimated Cost for ZAO Bot

  • FID registration: ~$0.30 one-time
  • Storage: ~$0.30/year
  • Neynar Professional: $49/mo
  • LLM API: ~$10-30/mo
  • Hosting: ~$5-20/mo
  • Total: ~$65-100/month

8. Best Practices for Community Bots

Do

  • Be helpful, not promotional: "Winning in niche channels is not about volume; it is about relevance, proof, and consistent helpfulness"
  • Respond to mentions only: Don't spam feeds with unsolicited posts. Trigger on @mentions and direct replies
  • Keep cast frequency reasonable: ElizaOS defaults to 90-180 minute intervals between autonomous posts -- this is a good baseline
  • Use dry-run mode first: Test all behavior before going live (FARCASTER_DRY_RUN=true)
  • Tag people and channels: Direct new users to relevant accounts and communities
  • Show expertise transparently: "State your role and scope when relevant"
  • Disclose bot nature: Make it clear the account is automated
  • Gate interactions: Use onchain criteria (token holdings, NFTs) to filter who can interact
  • Channel-specific behavior: Configure different response styles per channel

Don't

  • Don't auto-reply to everything: Only respond when tagged or when genuinely relevant
  • Don't over-automate relationships: "Automate logging and reminders, not relationship-building"
  • Don't treat Farcaster as a distribution channel: It's a "trust channel"
  • Don't spam likes/recasts: Keep MAX_ACTIONS_PROCESSING low (1 per cycle)
  • Don't exceed cast volume limits: Stay well under 1,000 casts/24h
  • Don't ignore context: Use memory systems to maintain conversation threading
  • Don't post generic content: Tie responses to exact thread context

9. Notable Failures & Anti-Patterns to Avoid

  1. Bot-inflated metrics: Farcaster's reported 40K-60K DAUs include significant bot activity; only ~4,360 Power Badge holders are genuinely active. Don't contribute to metric inflation.

  2. Generic engagement farming: Bots that like/recast everything get noticed and blocked. The community actively experiments with bot-exclusion methods.

  3. Shadow-banning risk: Despite decentralization claims, some users report being shadow-banned. Keep bot behavior within normal parameters.

  4. Ignoring storage economics: If your bot posts too frequently without enough storage units, casts get rejected. Monitor storage.

  5. Over-reliance on single provider: Neynar now controls both the protocol and API layer. Consider building with awareness of this centralization risk.

  6. Token-first approach: Agents that launch tokens without genuine utility (like many Clanker-spawned tokens) face community backlash. Build utility first.

  7. Ignoring Frames v1/v2 migration: Using outdated Frames v1 syntax will break. Always use Mini App patterns (fc:miniapp tags).

10. ZAO Bot Opportunity Analysis

Gap in the Market

No music-focused AI agent exists on Farcaster. Sonata is a music client/aggregator but is not an AI agent. Sound.xyz has integration but no bot. This is a clear opportunity.

What a ZAO Bot Could Do

  1. Music Discovery: Surface and recommend music shared in the /music channel and ZAO community
  2. Onboarding Assistant: Help new members understand ZAO, answer questions (like the Farcaster Support Agent does for Farcaster docs, but trained on ZAO content)
  3. Community Engagement: Welcome new members, highlight contributions, facilitate introductions
  4. Respect Token Integration: Because Farcaster accounts have linked wallets, the bot can check Respect token holdings and tailor interactions
  5. XMTP Bridge: Like the Farcaster Support Agent (accessible via XMTP), enable private conversations
  6. Event Notifications: Announce governance proposals, voting deadlines, new music submissions
  7. Curation: Help surface the best music submissions using community signals

Recommended Stack

  • Framework: ElizaOS with @elizaos/plugin-farcaster
  • API: Neynar Professional plan ($49/mo)
  • LLM: Claude or GPT-4o for response generation
  • Signer: Neynar-managed signer
  • Hosting: Same infra as ZAO OS (Vercel + background worker)
  • Knowledge Base: Train on ZAO research docs (research/ directory), community.config.ts, and Farcaster /music channel content

Implementation Priority

  1. Mention-triggered Q&A (respond when tagged with ZAO questions)
  2. Channel monitoring (watch /music for interesting shares)
  3. Scheduled community updates (governance votes, new members)
  4. Music recommendation engine (surface tracks based on community engagement)

Sources Index

Primary Documentation

Analysis & News

Code & Frameworks

Music on Farcaster