AgentLed CLI
Install AgentLed in the terminal and give your AI agent workflow tools, integrations, approvals, and business memory.
npx @agentled/cli setup
The setup command connects your local agent client to an AgentLed workspace.
Raw CLI documentation is also available at /docs/cli.md.
What your agent can do
Once connected, your agent can create workflows from natural language and operate them through MCP tools.
Outbound workflow
agentled create "Find fintech CTOs, enrich contacts, score fit, and draft outreach"
Research workflow
agentled create "Research competitors and summarize pricing changes"
Content workflow
agentled create "Generate five approved LinkedIn posts from recent signals"
Workflows compound
Each workflow can store useful context in the knowledge graph, so the next run starts with more business memory.
A common chain looks like:
- Detect signals. Find what changed in the market or account list.
- Enrich context. Pull the data needed to make a decision.
- Draft output. Generate the message, record, or report.
- Request approval. Pause sensitive work until a reviewer approves.
Use approval gates for anything that reaches a customer or production system.
Available MCP tools
The MCP server exposes workflow, knowledge, approval, and execution tools.
| Capability | Tools |
|---|---|
| Workflows | create, list, run, inspect |
| Knowledge graph | search, read, write |
| Approvals | queue, approve, reject |
| Executions | status, logs, retry |
Business and user memory
AgentLed separates durable business context from short-lived conversation context.
Business memory
Shared workspace context such as ICPs, scoring rubrics, and approval rules.
User memory
Personal preferences and working context scoped to the operator.
Pricing
CLI
Free
Install package and connect to your workspace.
Pro
β¬23.90/mo
2,000 credits for solo operators.
Teams
β¬86.90/mo
7,000 shared credits for teams.
Headless usage
For CI, servers, and advanced setup, see /docs/cli.md for environment variables and package options.