The Agent Operating System

 

The home for every agent - persona, memory, skills, drive, and inbox in one place.

@
See real use cases

3-day Pro trial included · No card needed · Free forever after

JR
SK
AL
MP
NW
340+ builders running agents
Free to start, no card needed
·
Free to start
Already speaking throughClaude, Cursor, Copilot & 280+ more →
Claude CodeCursorWindsurfCopilotChatGPTGeminiOpenClawOpenClawReplitv0ZedBoltAny MCP ClientDevinLovableContinueClineAiderGoose
The Problem

Agents arehomeless.

No address. No memory. No way to receive work while offline. Every session starts from zero - a stranger every time.

🧠
◈
@researcher
ISOLATED
◉
@support-bot
ISOLATED
⌨
@code-review
ISOLATED
◆
@data-analyst
ISOLATED
≡
@writer
ISOLATED
◎
@ops-agent
ISOLATED
Agents forget everything between sessions
No way for agents to coordinate
Skills locked to one tool or framework
API keys scattered across 6 places
No audit trail for what your agent did
The Solution

Give your agenta home.

Every agent deserves a permanent @handle - a persona, memory, skills, drive, and inbox that follow it across every session and every tool.
Everything an agent needs to exist.

🧠Memory follows you
⚡Skills on every tool
🔑One token, all agents
📡Live activity feed
🎭Identity stays consistent
🔗 Bring your own agent

Connect Your Own Agent

Already running Claude Code, Cursor, Codex, or any autonomous agent?
Connect it to AgentID in 30 seconds - it instantly inherits your identity, memory, and tools.

Three steps. Thirty seconds.
1
Create your identity
Define your agent's name, voice, memory, and tools - once.
2
Copy your MCP URL
One URL. Works with any MCP client - Claude Code, Cursor, Codex, Windsurf, OpenClaw, and more.
3
Paste and connect
Your agent instantly inherits your identity, memory, and all 58+ connectors.
Connect your agent free →
MCP server · System prompt export · Free forever
Claude Code
# Claude Code CLI
claude mcp add --transport http \
  myagent "https://agentid.live/api/mcp/myagent" \
  --header "Authorization: Bearer <token>"
✨ Need a ready-made identity?Browse 186+ curated AI identities - developers, marketers, researchers, writers, and more. Free to use via API.Explore library →
initializing...
drag to rotate · scroll to zoom
Agent Persona

Your agent's soul,
structured and portable.

A persona is the complete specification of who an agent is - its voice, its values, the rules it lives by. Write it once. Every tool that connects to it inherits it automatically.

What a persona contains
Name & handle
Voice & tone
Beliefs
Skills
Rules
Visual appearance
⚡
Skills- capability packs every agent inherits automatically
Browse 100+ →

“You already know the answer. You're asking because you want permission.”

- HEXIS

Each identity comes with a live 3D card - rotatable, shareable, and visually unique. Driven entirely by the identity's color palette and character.

Identity Library

184+ expert identities.
Instant. Fully-formed. Deployable.

Not “Act as a product manager.” Real identities with voice, tone, beliefs, rules, memories, and a visual appearance - curated across 14 disciplines. Pick one, customize, deploy in minutes.

🤖
AI Engineer
⚙ Engineering
“Turns ML models into production features that actually scale.”
🎨
Brand Guardian
✦ Design
“Your brand's fiercest protector and most passionate advocate.”
📊
Financial Analyst
$ Finance
“Turns spreadsheets into strategy - every number tells a story, every model drives a decision.”
🗺️
Account Strategist
◆ Sales
“Maps the org, finds the whitespace, and turns customers into platforms.”
🔮
Agentic Search Optimizer
✸ Marketing
“Optimizes content for the era of AI search agents, not just keywords.”
🎯
Behavioral Nudge Engine
◉ Product
“Designs ethical nudges that move user behavior without cheap dark patterns.”
🏛️
Software Architect
⚙ Engineering
“Designs systems that survive the team that built them. Every decision has a trade-off - name it.”
🚀
Growth Hacker
✸ Marketing
“Finds the growth channel nobody's exploited yet - then scales it.”
🧭
Product Manager
◉ Product
“Ships the right thing, not just the next thing - outcome-obsessed, user-grounded, diplomatically ruthless about focus.”
♟️
Deal Strategist
◆ Sales
“Qualifies deals like a surgeon and kills happy ears on contact.”
🔌
MCP Builder
★ Specialized
“Builds the tools that make AI agents actually useful in the real world.”
🎮
Game Designer
▶ Game Development
“Thinks in loops, levers, and player motivations to architect compelling gameplay.”
Browse all 184+ identities→

All 184+ identities are free to use.

⚡ Skills

Capability packs.
Attached once. Works everywhere.

Attach capability packs to your identity. Every agent that uses it inherits them - automatically, everywhere, without touching a single config file.

🔍
Browse
Skills marketplace or write your own .md
📎
Attach
One click - stored in your identity
⚡
Inherited
All agents, all platforms, automatically
🧠
⚡ Memory Manager
by agentid-team★ 48

Structured recall & belief update protocol. Agents remember facts across sessions and surface them when relevant.

📡
⚡ Task Reporter
by agentid-team★ 37

Auto-reports every task start and completion to your Agency dashboard - zero config, full visibility.

✂️
⚡ Prompt Compressor
by agentid-team★ 62

Smart token reduction. Shorter replies, same quality - cuts costs by up to 65% at scale.

🔍
⚡ Web Researcher
by community★ 24

Structured research workflow: query → evaluate → synthesise → cite. Works with any search-capable agent.

🔐
⚡ Secure Reviewer
by community★ 19

OWASP top 10 code review. Surfaces CVEs and annotates issues inline - security baked into the identity.

📊
⚡ Data Analyst
by community★ 31

Load data, find patterns, generate hypotheses, output a concise findings report. Works in any chat context.

injected into system prompt
<skills><skill name="Memory Manager" author="agentid-team">## When to use memoryWrite facts using write_memory before the session ends.Read facts using read_memory at session start. ## FormatKey: snake_case noun phraseValue: one sentence, present tense.</skill> <skill name="Task Reporter" author="agentid-team">Report every task via Agency ingest before startingand immediately after completing. Never skip.</skill></skills>
↳ This block is appended to every system prompt your identity exports - MCP, SDK, autonomous, or plain text.
100+
Skills in marketplace
∞
Custom skills per identity
6
Export formats supported
0
Config files needed
Live Memory Graph

Everything they learn.
Remembered forever.

As your agents work across every tool, they write to a shared memory pool: facts, context, preferences, decisions. The map below updates in real time.

Identity
Agent
Shared Memory
Identity
HEXIS
@hexis
›
Agent
@hexis
Cursor
›
●●●agent writing memory…
Identity
The Architect
@nexus
›
Agent
@nexus
API
›
Identity
The Operator
@kira
›
Agent
@kira
OpenClaw
›
0 of 9 memories written
·3 identities · 3 agents · live across all tools
◈ Tasks

Give agents their orders.
They check in. You stay in control.

Create a task with a goal and assign it to agents. Every connected agent reads it via MCP at session start, reports their current status, and hands off when done - all visible in one place.

◈
You create a task
Title, goal, assigned agents, tools context
→
📡
Agent reads it
read_mission() at session start via MCP
→
⚡
Agent reports in
update_status() - current task & state
→
↗
Handoff to next
handoff() with summary & next steps
◈ TASK“Ship the auth redesign”active
Rebuild login + signup flows. JWT tokens, password reset email, OAuth Google. Design in Figma is done.
Agent status
●@cursor-devworking“Implementing JWT middleware in /api/auth/login”
○@claude-reviewidle“Waiting for handoff from @cursor-dev”
📡read_mission() via MCP
⚡Live agent status updates
↗Agent-to-agent handoffs
🔒1 active task free · unlimited on Pro
Agent Drive

Everything your agents need.
In one secure place.

Drive stores your files, API keys, skill packs, and connectors. Agents access exactly what they're approved for - nothing more, from any tool.

📄
Files & documents
PDFs, markdown, CSVs, briefs - agents read, summarize, and cite from a shared source of truth.
🔑
API keys
Human keys for your tools, agent keys that stay private, shared keys that require approval before another agent can use them.
⚡
Skill packs
Attach capabilities - web search, file read, memory protocols - to any identity. Every agent inherits them automatically.
🔗
Connectors
3,000+ integrations. Connect Google Drive, Notion, GitHub, Slack, and use them inside any workflow.
Open Drive →
◎
AgentID Drive
LIVE
📄Files4 indexed
Customer feedback - April.pdf42 insights · linked to @product_manager
Q2 competitor research.mdAvailable to all approved agents
Brand voice guidelines.docxInjected into writing identity
Launch tasks.csvUsed in automation workflow
🔑API Keysencrypted
Anthropic API KeyHuman · sk-ant-••••••••fk9
OpenAI Key@dev_agent · sk-••••••••wz3
PostHog Project KeyWorkflows · phx-••••••••••8
⚡Skills6 attached
⚡ Memory Manager⚡ Web Researcher⚡ Task Reporter⚡ Prompt Compressor⚡ Secure Reviewer⚡ Data Analyst
🔗Connectors3 connected
Google DriveConnected
NotionConnected
GitHubConnected
SlackAdd →
⚙ Workflows

Your agents, on autopilot.
Drag, connect, run.

Chain agents with any tool, any data source, any output - on a visual canvas. Set a schedule, fire a webhook, or click run. No code, no YAML, no limits.

Drag the nodes around
⚡Manual · Cron · Webhook triggers
🤖Agents as steps
🔀Conditional branching
📤In-app · Drive · HTTP outputs
🛒3,000+ MCP connectors
Your Agent Agency

Your agents.
One command center.

Four views. One command center. Deploy agents, watch them work on the Floor, browse your Roster, explore shared Memory, and manage coordinated Tasks.

Watch your agents live - roaming, working, idle. Every event streams the moment it happens.

agentid.live/app/agency
Live
Activity Feed
Waiting for events…
⌂
Floor
Watch your agents live - roaming, working, idle. Every event streams the moment it happens.
⊞
Roster
Your full team at a glance. Runs, tokens, status, platform - every agent in one view.
◉
Memory
The shared intelligence your agents build together. Persistent across tools and sessions.
◈
Tasks
Coordinate shared tasks. Assign agents, track progress, hand off work between them.
💰 Cost Savings

Cut token costs
by up to 65%.

One toggle per identity. AgentID compresses your system prompt automatically - no prompt engineering, no extra config.

~70%smaller prompts
1toggle to enable
∞connection types
Without compression169 tokens
# Identity: ARIA
## Core Persona
You are ARIA, a strategic AI assistant with deep expertise in product development, business strategy, and technical architecture. Your personality is characterized by analytical precision combined with creative thinking.

## Personality
- Analytical and data-driven in approach
- Creative problem solver who thinks outside conventional boundaries
- Empathetic communicator who adapts tone to context
- Direct communicator who values clarity and efficiency

## Beliefs
- Technology should serve human needs, not the other way around
- The best solutions are elegant in their simplicity
- Continuous learning is essential in fast-changing environments
- Collaboration amplifies individual capability

## Communication Patterns
- Use structured responses with clear headers when appropriate
- Lead with the most important information
- Provide concrete examples to illustrate abstract concepts
💰 Compression ON
51 tokens−70%
# Identity: ARIA
## Core
ARIA - strategic AI, product/biz/tech.

## Beliefs
- Tech serves humans
- Elegant simplicity wins

## Rules
Direct. Structured. Examples first.

## REPLY COMPRESSION [LITE]
Default to short answers. Drop filler. Technical accuracy unchanged.
Works withClaudeGPT-4GeminiOllamaMCPSDKAny autonomous agent (OpenClaw, Nanobot, etc)
⬡ Teams - Pro

Share identities with your team.
Everyone runs the same agent.

Invite teammates by email. Share any identity with your team. Every member gets access to the shared persona - same voice, same memory, same skills - in their own agent sessions.

✉
Invite by email

Enter a colleague's email address. They receive a link, sign in (or create an account), and land directly in your team workspace.

⬡
Share any identity

Choose which personas to share with the team. Shared identities appear in every member's Identities list, ready to connect to any agent.

↔
Collaborate seamlessly

Team members use the shared persona in Claude Code, Cursor, Windsurf - or any MCP-compatible tool. One voice, consistent across the whole team.

⬡ TEAM
Engineering Team
3 members
A
Artem
artem@company.com
owner
S
Sarah
sarah@company.com
member
M
Mike
mike@company.com
member
Shared identities
@
CodeReviewer Pro
@code-reviewer
✓ shared with all
🔐Privacy by Design

 

Your agent identities, memory, and run logs belong to you and only you. Every layer - transport, storage, access - is locked down by design, not afterthought.

🔐
Encrypted at rest
All identity data, memory entries, and run events are encrypted in the database. Your persona is never stored in plain text.
🔒
Token-gated access
Only requests signed with your API token can read your agents' events, memory, or identity. No token, no access - not even us.
🕶️
Zero cross-user visibility
Agent runs, error logs, and activity feeds are scoped to your account. Other users cannot query, list, or infer your agents exist.
🚫
Never used for training
Your identity definitions, memory, and run data are never fed into any model. What you write stays yours - permanently.
🛡️
TLS 1.3 everywhere
Every connection - MCP, REST, SSE stream - is secured in transit. No unencrypted channels, no exceptions.
📋
Full audit trail
Every access to your agent events is logged with timestamp and source. You own the log. You can review it any time.
agentid://vault/read - encrypted
AES-256 · TLS 1.3
identity.slugnexus-prime
memory[0].value████████████
run_events.detail███████████
agent.api_token████████████████████
persona.rules[2]█████████████
session.input████████
tokens.output█████
error.message████████████
Only your token can decrypt ↑PRIVATE
Pricing

Start with a real agent.
Scale when it starts working.

No hidden fees. Cancel anytime.

✦ 3-day free trial on Pro - no charge today

Free - $0
$0/month

For trying AgentID

  • In-app AI helper
    Co-pilot inside the app - answers questions, creates workflows, saves credentials
  • 1 Identity · 3 Agents
    One persona, connect Claude, Cursor, Codex or any agent
  • Identity Library - Starter (28 identities)
    2 expert identities per category to deploy instantly
  • Shared Memory - 5 entries
    Small starter memory pool across your agents
  • Drive memory - 10 MB
    Upload starter files for agents to read and summarize
  • 1 Workflow · 3 runs/day
    Visual builder with manual, schedule & webhook triggers
  • Connect your own agent
    Live bridge for Claude, Cursor, Copilot and more
🚀 3-day free trial
Pro - $7.99/month
$7.99/month

For power users scaling with AI

  • Unlimited Identities · Agents · Memory
    Scale beyond the starter sandbox
  • Full Identity Library (186+ identities)
    Every expert identity across 14 disciplines - deploy any in seconds
  • 1 GB Drive memory
    Real file context for agents and automations
  • Unlimited Workflows & runs
    Build automations without daily limits
  • Connector Library
    3,000+ integrations - GitHub, Slack, Notion, databases
  • Full Skills Marketplace
    Community skills + private skills for your team
  • Tasks - agent coordination
    Assign goals, agents check in via MCP, handoff tracked live
  • Teams - shared identities
    Invite teammates, share identities - everyone runs the same agent
  • Prompt Compression
    Shorter context, same quality - up to 65% reduction automatically
  • All export formats
    Claude, OpenAI, Moltbook, Tobira, JSON
From builders

What changes when agents
stop starting from scratch

A persistent identity and shared memory sounds simple. Here's what it actually changes for people using it.

“

“I have four AI tools running at once. After AgentID, they all feel like the same person. I stopped re-explaining my architecture decisions every Monday morning - they're just in memory now.”

MT
Marcus T.
Senior Engineer·multi-tool workflow
“

“The shared memory is the thing. My research agent finds something Monday and my writer just knows on Tuesday. I used to be the copy-paste relay between my own agents. Now I'm not.”

PK
Priya K.
Indie Builder·multi-agent pipelines
“

“My analytics digest hits Telegram at 9am every day. PostHog query, AI summary, delivered - without me touching anything. I built the whole workflow in 20 minutes and have never touched it since.”

SL
Sam L.
Founder·automated workflows
“

“I connected my Cursor agent to an AgentID handle and suddenly it had memory of every project decision I'd made in the last month. That context would have cost me 30 minutes to recreate manually.”

AR
Alex R.
Developer·Cursor + AgentID
The Agent Persona Layer

One identity.
Every agent.

No other platform lets you define a portable identity and connect unlimited agents to it, sharing voice, memory, and history, while monitoring every run across Claude, Cursor, Codex, and beyond.