Articles for people who
ship the stack
Rewrites on React, Node.js, TypeScript and AI. A translated body is shown when it exists. If it does not, you see the English text and a note that says so.
Tagged: Human-in-the-Loop
LangGraph in Python: State, Parallel Nodes, Checkpoints and Human Review
A hands-on LangGraph tour in Python: TypedDict state and reducers, edges, checkpointers, interrupts, and an email triage workflow with human approval.
25 min readParallel Coding Agents With Git Worktrees, Conductor and Antigravity 2.0
Parallel coding agents with git worktrees, Conductor and Antigravity 2.0: a coordinator, one worktree per agent, and a dozen SRE features in about five hours.
14 min readHuman-in-the-Loop for AI Agents: Approval Gates, Escalation, Audit Trails
Where an agent needs human sign-off, four patterns (gates, escalation, collaborative edits, sampling), pause and resume with LangGraph, timeouts, audit logs and pitfalls.
18 min readBuild ShopBot: a LangGraph Support Agent With Memory, Tools and Refund Approval
Build a customer support agent in LangGraph module by module: state, tools, nodes, routing, rolling summaries and an interrupt that waits for refund approval.
18 min readThree LangChain Middleware Guards That Keep Agents From Leaking Secrets
Stop an agent from sending API keys to the model or deleting files unasked: PIIMiddleware, HumanInTheLoopMiddleware, call limits and why middleware order decides safety.
10 min readLangGraph Human-in-the-Loop: interrupt(), State Edits and Time Travel
LangGraph human in the loop: approve tool calls with interrupt(), add feedback gates, edit state by hand and rewind with time travel, with working examples.
12 min readBuilding a DeepAgents Harness: Permissions, Subagents and Human Approval
Building a DeepAgents harness for a system-design planner: file backends, access control rules, subagents, human approval gates, skills and MCP tools.
24 min readGate Agent Actions by Effect, Not by Verb: Lessons From a Five-Agent Swarm
How a small multi-agent setup routed around a keyword-based approval gate, what the agents built on their own, and why permissions must describe effects, not words.
14 min readSelf-Repairing Automation Without an Allow-Clicker: Gates That Work at 3am
Design unattended auto-repair for scheduled agent routines: an outside watchdog, actions typed by reversibility, canary rollouts and status reports backed by evidence.
6 min read
About these articles
Request a 24h estimate
Need the same stack in a production operator layer? Send the brief — estimate within 24 hours.