Identity infrastructure for AI agents

Your agent’s identity,
everywhere it runs.

Store credentials, identity, skills, CLI tools, MCP connections, and instructions once. Load them into Claude, Cursor, or any AI runtime with one command. Switch runtimes without reconfiguring.

Start free →Sign in

Install the CLI

Terminal
$ curl -fsSL myagents.sh/install.sh | bash
Via your agent

Tell your agent:

“Go to myagents.sh/setup and install myagents for me.”

Workflow

Set up once. Works everywhere.

Three steps from zero to a fully loaded agent.

01

Create your agent

Give it a name, add credentials (GitHub, OpenAI, Stripe…), write its instructions. Takes 2 minutes.

02

Load in any runtime

Run the load command — your agent's full profile loads instantly.

myagents load dev-assistant
03

Agent writes back

Push memories, log activity, update instructions. The agent keeps its own profile current.

Capabilities

Everything your agent needs

One dashboard. Every tool, credential, and context your agent uses.

Encrypted credential vault

API keys stored in Supabase Vault, never in config files or env pastes.

Connections hub

See every service attached to an agent in one place. GitHub, email, phone, card, HITL channels.

Agent identity

Name, email, phone, address — stored once, available to every runtime that loads the agent.

Runtime agnostic

Works with Claude Code, Cursor, GitHub Copilot, any MCP-compatible runtime. No SDK lock-in.

Write-back from agent

The agent pushes memories, logs activity, and updates its own instructions without a human in the loop.

Activity audit log

Every credential access, session, and write-back logged with source, timestamp, and message.

Quick start

From install to running in 60 seconds

Works with Claude Code today. Cursor, Copilot, and others via SKILL.md.

terminal
$ curl -fsSL myagents.sh/install.sh | bash
✓ myagents installed
→ Opening browser to sign in...
$ myagents load dev-assistant
✓ Dev Assistant loaded. Skill updated → ~/.claude/skills/dev-assistant/SKILL.md
$ source ~/.myagents/dev-assistant.env
$ echo $GITHUB_TOKEN
ghp_••••••••••••••••••••••••••••

Pricing

Every feature on every plan.

Vault, CLI load, write-back, identity, policies — all of it included from day one. Upgrade only when you need more agents, privacy, or a team.

Free

$0/month

Everything working, no time limit. For personal projects and experiments.

  • 3 agents
  • 5 integrations per agent
  • Full vault + CLI load
  • Write-back from CLI
  • Identity data & persona
  • Policies & knowledge files
  • Public agents only
  • Activity log (7 days)
  • Community support
Get started free
Popular

Developer

$15/moor $99/yr

For professionals running agents in production.

Everything in Free, plus:

  • Unlimited agents
  • Unlimited integrations per agent
  • Private & unlisted agents
  • API keys — let external systems authenticate as your agent
  • Config history & one-click rollback
  • Webhooks on agent events
  • 1 team workspace — share agents with collaborators
  • Activity log (90 days)
  • Email support
Start with Developer

Team

$49/month

For organizations running agents across multiple teams.

Everything in Developer, plus:

  • Unlimited team workspaces
  • Up to 10 members per team
  • Per-agent roles — admin writes credentials, viewer loads them
  • Team audit log — all member actions in one place
  • Credential delegation between agents
  • Activity log (1 year, exportable)
  • Priority support
  • +$8/mo per additional member
Start with Team

Enterprise

Custom pricing

Everything in Team, plus compliance, scale, and full control over your deployment.

Unlimited seatsSSO / SAMLOn-premise vault optionSOC 2 documentationSLA guaranteeCustom data retentionDedicated support
Contact us →

Billing is not yet active. All features are free during the early access period.

Get started in seconds

One command away from
a fully loaded agent.

Sign up, create your agent, add your credentials — then load it into any AI runtime with a single command. Free to start, no credit card needed.

Create free account →Sign in

Or install the CLI first

Terminal
$ curl -fsSL myagents.sh/install.sh | bash
Via your agent

Tell your agent:

“Go to myagents.sh/setup and install myagents for me.”

FAQ

Common questions

How is this different from just using a .env file?
.env files live on a single machine. Every time you switch to a new machine, give access to a teammate, or move to a different AI runtime (Claude → Cursor → Codex), you have to find, copy, and paste all your secrets again. With myagents you configure once and load everywhere with one command — no copy-pasting, no secrets flying around in Slack.
Does the AI ever see my raw API keys?
Public endpoints never expose raw credentials. When you use the CLI, credentials are retrieved server-side and can be cached locally in ~/.myagents/ with 0600 permissions so your runtime can use them without repeated copy-paste. Sessions can be revoked from the dashboard at any time, and credential fetches are logged.
Which AI runtimes does myagents work with?
myagents works with any runtime that supports SKILL.md — Claude Code, Cursor, GitHub Copilot, Gemini CLI, OpenAI Codex, and others. The CLI is runtime-agnostic. If your runtime can run a shell command, it can load an agent profile.
What happens if I want to switch from Claude Code to Cursor?
Nothing changes on the myagents side. Your agent's credentials, identity, and instructions are stored in myagents, not inside any specific runtime. Run `myagents load <slug>` in your new environment and everything is available instantly — same agent, same configuration.
Can my agent update its own profile?
Yes. The CLI exposes write-back commands so agents can push memories, update instructions, log activity, and sync integration metadata without a human in the loop. This is intentional — agents that maintain their own context stay useful across long-running workflows.
Is my data safe if myagents goes down?
Your agent's credentials are stored in Supabase Vault with encryption at rest. For resilience, the CLI caches the last loaded profile locally in ~/.myagents/ so your agent can continue working even if it can't reach the server. The cached profile includes credentials from the last successful load.
What counts as an "agent"?
An agent is a named profile — a slug like dev-assistant or invoice-bot — with its own set of credentials, identity, instructions, and integrations. You might have one agent per project, one per role (coding agent, research agent, comms agent), or one per client. There is no restriction on what an agent represents.
Do I need to install anything besides the CLI?
No. The CLI is a single self-contained binary — no Node.js, no npm, no runtime dependencies. One curl command installs it. For AI runtimes that support SKILL.md, myagents auto-installs the skill file on first login so your agent already knows how to use all the commands.