Cargo x Claude Code

Claude Code is the agent.Cargo is the production runtime.

Install the Cargo with one prompt in Claude Code, then run your first workflow in minutes.

Start building for free

The full stack

Claude Code is one layer. Cargo is the layers underneath.

The reasoning layer needs a runtime underneath: data, execution, observability, governance, and integrations.

system stacklayers live
Reasoning & skillsClaude Code
Tools, MCP & APIs150+ providers
Unified GTM data layercompanies · contacts · ops
Executionretries · rate limits
Observabilityevery run logged
Governanceaudited writes
What Cargo adds

The structural gaps Claude Code can't close alone.

Three things you need to take an agent from a sharp prompt to a system you can put in front of revenue.

  • Unified data layerA live, structured layer of companies, contacts and opportunities the agent can read from and write to.
  • Observability & governanceEvery routing decision and CRM write is logged, retryable, and protected by field rules and approvals.
  • Maintained integrations150+ enrichment providers and CRMs maintained for you, so API churn and rate-limit shifts do not break runs.
Quickstart

Sign up for Cargo and get started in 2 steps.

Watch the walkthrough, then run the same flow in your terminal.

Watch on YouTube
01Install Cargo CLI and skills

# Paste this prompt in Claude Code

Install the Cargo CLI and skills by following every step in https://api.getcargo.io/agent-install.txt

02Prompt your use case
claude-codeLookalikes of closed-won deals
>Using cargo-skills - build and deploy a play on the HubSpot Deals model. Filter to closed-won deals only. Find lookalike companies via Ocean.io searchCompanies seeded by the deal domain. For each lookalike, resolve its LinkedIn company ID and find Engineering Manager leads via Sales Navigator. For each lead, dedup contact and company against HubSpot and stop if either exists. Enrich company and contact, then upsert both to HubSpot with key properties and a cargo_campaign_tag. Validate, deploy, commit.
⏺Write(engine/campaigns/closed-won-lookalikes-nodes.json)
⎿Removed: "description": "only process closed-won deals"
⎿Removed: "name": "Map deal domain"
⎿Removed: "description": "map deal domain as seed for lookalike search"
⏺Bash(cargo-ai run closed-won-lookalikes)
⎿{ "batch": { "status": "finished", "stats": { "runs": 142, "upserts": 89, "duplicates_skipped": 53 } } }
Need help?

Book a call

Share your use case and we'll help you get your first workflow running.

Book a call
Observability & cost

Every action traced. Nothing runs blind.

Claude Code executes. Cargo tells you exactly what happened — which node fired, what it returned, what it touched downstream.

Last run · enrich and route
run · 8b2f1
  1. Trigger
    42ms
  2. Find records
    116ms
  3. Enrich companyretry 1/3 · provider 502
    1.4s
  4. Update CRM
    612ms
3 ok/1 retry/0 fail
✓ 2.4s total
  • Per-node success rates & error logsKnow exactly which step broke and why — not just that the workflow failed.
  • Automatic retry with configurable backoffSet max attempts, initial interval, and backoff coefficient per node. No silent failures.
  • Field protection on every CRM writePrevent agents from overwriting critical fields. Skip-if-exists logic. Approval flows before anything irreversible runs.
  • Credit tracking by workflowKnow exactly what each play, tool, and agent costs. No billing surprises at scale.
Ship with Cargo

Give your GTM workflows a runtime to ship on.

Run from terminal with Cargo CLI, then monitor results, governance, and execution in Cargo.

Start building for freeRead the CLI docs