The Infrastructure Layer for the Claude Agent SDK

MCP: tool bridge
50: MCP tools
5: SDKs, one layer
1 cmd: docker compose up

agentFast is the production layer for the Claude Agent SDK. The Claude Agent SDK gives you a capable agent loop with tool use and MCP. agentFast adds the rest of what a real deployment needs — through the same store seam that backs every other agentFast adapter.

What agentFast adds to the Claude Agent SDK

An MCP tool bridge

The 50 MCP-native tools, exposed idiomatically behind the same fixed 12-code error vocabulary and guardrails as every SDK.

Durable, crash-proof runs

Checkpoint-per-iteration to Postgres, so a run survives a killed process and resumes from the exact step.

Memory

Short-term buffer, long-term vector recall, episodic log and scratchpad — with compaction for long runs.

Observability

A full step tree with per-step cost, latency and prompt-cache accounting, plus OTEL and LangSmith export.

Guardrails

PII redaction before the model sees it, injection defense, abuse blocking and rate limits.

Durable human-in-the-loop

Gate risky tools; each approved action runs exactly once via a database compare-and-set.

From clone to a running Claude Agent SDK agent

1
Install with the Claude Agent SDK extra
2
Scaffold an agent
3
Bring up the full stack
4
Gate on quality

bash

pip install -e ".[claude-agent-sdk,dev]"
agentfast new support --sdk claude_agent_sdk
cd support-agent && docker compose up
agentfast eval

The SDK vs. agentFast on the SDK

	Claude Agent SDK	agentFast on the SDK
Agent loop + tool use + MCP	✓ its core strength	Uses the SDK directly
Memory (vector + episodic)	— bring your own	✓ Included
Observability + cost tracing	— bring your own	Full step tree
Guardrails (PII, safety, limits)	— bring your own	✓ Included
Durable human-in-the-loop	— bring your own	Survives a crash, exactly-once
Evals in CI	— bring your own	✓ Included
Working reference agent	Examples	SupportAgent, point it at your KB

One layer, five SDKs

The whole point of agentFast is that the production layer is SDK-agnostic — the seam holds across a vendor-CLI-driven model exactly as it does a hand-rolled one. Explore observability, guardrails, and durable execution.

Read why the production layer is the part every team rebuilds, or start from a working support agent.

Stop rebuilding the plumbing.

agentFast is the production layer — memory, 50 tools, observability, evals, guardrails, human-in-the-loop — for LangGraph, CrewAI, the Claude & OpenAI Agent SDKs & Vanilla. Own it for life.

Get agentFast — from $89

Keep reading

The Production Layer Every AI Agent Needs (and Why LangGraph Doesn't Give It to You)

9 min read

LangGraph vs CrewAI vs a Vanilla Loop: Which Should You Actually Use in 2026

11 min read

How to Add Human-in-the-Loop Approval to Any AI Agent (With Durable Execution)

10 min read