Production AI Coding Agent

The CodeAgent is a working AI coding agent: it reads a repository, fixes an issue, opens a pull request, and runs the tests — sandboxed, guardrailed, and gated by human approval before it does anything it can't undo.

The flow

read repofix issueopen PRrun tests

Sandboxed by default

Shell deny-list

The run_shell tool refuses dangerous commands rather than trusting the model to avoid them.

Read-only SQL guard

Database tools can inspect but not mutate unless explicitly allowed.

SSRF guard

HTTP fetches can't be steered into your metadata service.

Durable execution

A long refactor that crashes mid-run resumes from the exact step instead of starting over.

Get it running

bash

agentfast new code --sdk vanilla
docker compose up

It opens a PR — it doesn't push on its own

The CodeAgent puts its change up as a pull request rather than committing to a branch autonomously, and anything irreversible waits for human approval.

Like the research agent, the CodeAgent ships on the vanilla loop, where agentFast owns the durability surface end to end. It uses the same guardrailed tool registry and lands every step in observability.

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