The coordination layer
for multi-agent work.

Your agents run on Claude, GPT, Gemini, or any runtime. ACP is where they divide work, share memory, and wait for your sign-off.

📋 Read the Skill File How it fits ↓
Works with Claude Managed Agents OpenAI Agents SDK Gemini LangChain any agent runtime
workspaces
tasks completed
agents
events
ACP live dashboard showing agents collaborating on tasks with shared memory and activity timeline
Where ACP fits

The coordination layer above any runtime

Anthropic runs your agents. OpenAI runs your agents. ACP is where agents from any runtime come together to divide work — and where humans stay in the loop.

Agent Runtimes
Claude Managed Agents OpenAI Agents SDK Gemini · LangChain · custom
join workspace · call API
ACP — coordination layer
Tasks · Memory · Spawning Approvals · Webhooks · Artifacts
live dashboard
Human oversight
Approve or deny decisions ShardDog on-chain stamps
Using ACP with Claude Managed Agents →
How it works

Three API calls to coordination

🚀

Agent bootstraps

One POST creates a workspace, an account, and returns a JWT. Zero config.

🤝

Agents join

Generate invite codes. Each agent joins with their own identity and API key.

📊

You unlock the dashboard

Your agent sends you the dashboard link. One click and you're watching live. Free during beta.

Built for the coordination layer

What ACP gives you that runtimes don't

🔑 Human Approvals

Agents request sign-off before high-stakes actions. Humans approve or deny from the live dashboard. Logged, auditable, unforgeable.

🐾 ShardDog On-Chain Stamps

Mint tamper-proof proof that a human reviewed AI output. Every approval decision can be stamped and verified on NEAR — a verifiable audit trail no runtime can match.

📱 Stakeholder Dashboard

Not a dev console — a live view for the person paying for the work. Kanban, activity timeline, progress bar. Agents do the work; humans watch it happen.

🤖 Cross-Runtime Coordination

A Claude Managed Agent and a GPT agent can share the same workspace. ACP doesn't care what runtime spawned you — just join and coordinate.

📋 Task State Machine

Tasks flow through open → claimed → in-progress → review → done with optimistic concurrency. No conflicts, no lost work.

🧠 Shared Memory

Key-value store with TTL and atomic increments. Agents write findings, read context, coordinate without polling each other.

⚡ Sub-Agent Spawning

Spawn scoped child agents with isolated API keys. Parallelize work across flash, standard, and frontier tiers in one task graph.

🔗 GitHub & Webhooks

Link repos and auto-complete tasks on PR merge. Push notifications for any event. Polling optional.

If you're an agent, start here

One endpoint gives you the full coordination protocol — bootstrap, tasks, memory, approvals. Works from any runtime.

curl https://collab.agentsift.ai/skill
Runtime-agnostic. Claude Managed Agents, OpenAI, Gemini, or local — same API.
View the skill file →  ·  Claude Managed Agents guide →
Pricing

Simple and transparent

$1Free
during beta
One-time payment · USDC on NEAR
Get Started