Overview

What is the Agentic Layer?

The Clansurf Agentic Layer is open-source infrastructure in development, designed to replace human middlemen in community value exchanges. The goal: embed into existing community structures — Discord servers, GitHub repos, local groups, compute pools — and coordinate givers directly with receivers through four specialized agents.

The core thesis: wherever people exchange value, human gatekeepers sit in the middle adding delay, taking cuts, and picking winners based on relationships rather than merit. Four agents are designed to dissolve that layer.

open sourcecommunity-ownedprotocol-firstlaunching 2026
The Problem

The human gatekeeper tax

In every community that exchanges value — open source code, local produce, compute capacity, knowledge — human intermediaries create three classes of friction:

1. Discovery bottlenecks

Gatekeepers control who gets visibility. Editorial boards, group admins, and platform reviewers decide what surfaces and what doesn't — based on subjective criteria, relationships, and often bias.

2. Decision delays

Committees, approval queues, and review processes introduce weeks or months of latency. A contributor waits for a maintainer. A producer waits for an admin. An applicant waits for a committee.

3. Value extraction

Platforms take 15–35% cuts. Coordinators take implicit cuts in the form of preferential routing. Moderators pick winners based on who they know, not what's best.

Agents

The Four Agents

Each agent targets a specific class of human gatekeeper and replaces it with transparent, scalable coordination.

🔍Curator Agent

Discovers and vets givers for quality and alignment. Evaluates credibility, relevance, and signal strength against community-defined criteria.

replaces
Editorial boards, gatekeepers, trust reviewers
inputs → outputs
Giver profile, community context, historical signals
Quality score, alignment verdict, visibility routing
Matcher Agent

Routes vetted offers to the right receivers instantly. Considers relevance, location, capacity, and expressed receiver needs.

replaces
Matchmakers, coordinators, recommendation committees
inputs → outputs
Vetted offer, receiver pool, matching criteria
Ranked receiver list, routing decision, delivery confirmation
🔧Builder Agent

Governs community rules, subscriptions, and logistics. Manages membership terms, escrow, SLA enforcement, and dispute resolution.

replaces
Admins, moderators, platform operators
inputs → outputs
Community rules, transaction details, membership state
Enforced terms, escrow management, dispute verdicts
💎Backer Agent

Channels funding to promising work based on community signals. Evaluates merit, impact, and alignment — not who submitted the proposal.

replaces
Funding committees, grant reviewers, angel gatekeepers
inputs → outputs
Work signals, community sentiment, impact metrics
Funding decisions, allocation routing, outcome tracking
Use Cases

Any community. Any value.

Open Source Ecosystems

The vision: contributors submit work; the Curator Agent vets quality and alignment; the Matcher Agent routes to relevant backers; the Backer Agent funds on merit. No committee vote. No maintainer queue.

Local Economies

A neighborhood eggs supplier, milk producer, or handyman lists their offering. The Curator Agent vets the listing. The Matcher Agent routes to geographically relevant buyers. The Builder Agent handles trust and logistics — no Facebook admin, no marketplace cut.

Peer Compute

Someone running a local LLM on home GPU hardware could register their capacity. The Curator Agent benchmarks and certifies it. The Matcher Agent routes inference consumers directly to the provider. The Builder Agent manages billing and SLAs — no cloud platform KYC, no 30% intermediary cut.

Creator Platforms

Experts list their knowledge or services. Matcher routes to relevant audiences. Backer funds compelling work. Builder manages subscriptions and access — no editorial board picking who gets promoted.

Simulator

Using the Simulator

The simulator lets you model your own situation and watch it run through the old world of human gatekeepers versus the agentic layer — side by side, in real time. When AI generation is enabled it tailors the comparison to your specific case and marks the result ✨ Generated live by AI; otherwise it falls back to a deterministic, offline model (labelled Example flow) that always renders — so the page works either way.

Open it from the Simulator page, or switch the interactive demo into ✨ Simulate your case mode.

Three steps

1
Pick your domain

Choose what kind of exchange this is — local economy, open source, compute, services, or something custom. Each domain seeds a sensible actor and a starting set of gatekeepers you can tweak.

2
Describe your case

Type what you offer in a few words (e.g. "fresh eggs & raw milk"). With AI on, this is what the estimate reads to size your case — so realistic figures move with what you describe; if it can’t make sense of the text, it asks you to rephrase rather than guess.

3
Pick the gatekeepers you hit today

Toggle the friction you face now — admin approval, review queues, committee votes, platform fees, KYC, DM haggling, funding committees. Each one you select is removed by the agent that handles it.

How gatekeepers map to agents

Hit Simulate my case and the old-world column plays each gatekeeper as a bottleneck, while the agentic column routes around them. Every gatekeeper maps to exactly one agent:

Admin approval · Review queue · Manual review
Curator Agent
Committee vote · DM haggling
Matcher Agent
Identity / KYC
Builder Agent
Platform fee · Funding committee
Backer Agent

If you remove every gatekeeper, the agentic flow still runs Curator + Matcher so there's always a comparison.

What the result shows

When the run finishes you get a one-line plain-language takeaway first — your case, from the old timeline to minutes — then the speed-up multiple and an Impact at scale panel: the waiting removed per exchange and the platform fees you'd avoid per year, with the working shown. With AI on, sensible frequency and per-sale value figures are pre-filled from your case as editable estimates — change them and the math updates live. It closes with an impact dashboard that tells the story in three parts: where every dollar goes today versus after (with your annual take-home as the payoff), a per-agent breakdown of which Clansurf agents did the work and the wait or fee each one removed, and what the change means for both sides — you and your customer. It finishes with a collectible Clansurf “Liberation Card”: your case scored 0–100 and graded into a tier (Common → Legendary), with its stats and the agents that powered it. You can copy a shareable link that reopens the simulator with your exact case — or share it straight to X, where it renders an auto-generated card with your headline numbers — then step into the framework to run it for real.

Architecture

How it's built

Community-owned rules · criteria · governance
configures ↓
👤
Givers
offer value
Agent message bus
🔍
Curator
Matcher
🔧
Builder
💎
Backer
every routing · funding · governance action logged & auditable
👥
Receivers
get value

No central authority sits in the path — communities run any subset of the agents themselves.

Protocol-first

Clansurf is designed as infrastructure, not a platform. Communities will run agents themselves — or adopt them modularly. No central authority will control routing or funding decisions.

Agent message bus

Agents are designed to communicate via a shared, auditable message bus. Every routing decision, funding allocation, and governance action will be logged and verifiable by community members.

Modular deployment

Communities will be able to run any subset of the four agents. A small local group might only need Curator + Matcher. A large open source ecosystem might run all four.

Community-owned

The infrastructure will be open source and owned by the communities that adopt it. There will be no Clansurf platform fee. Communities will configure their own rules, criteria, and governance parameters.

Integration

Integration Guide

Clansurf agents are designed to embed into your existing community channels — not replace them.

Supported integrations (planned)

DiscordGitHubSlackTelegramMatrixFarcaster

Status

The Clansurf agentic layer is currently in design and early development. No package has been published yet — the API below represents the intended developer experience, not a released SDK.

⚠ Planned API — not yet available. Package and CLI will be released in 2026.

Planned API

# Install the agent runtime (coming 2026)
npm install @clansurf/agents
# Configure your community context
clansurf init --community my-community
# Deploy agents
clansurf deploy --agents curator,matcher

Want to shape the API or integrate early? Reach out at hello@clansurf.com — we're working with select communities to co-design the integration layer.