Explore

Hire people who can't stop themselves from building

Josh Woodward describes his hiring approach: obsess over finding people who just can't stop themselves from building. Look for intellectual curiosity — reading stuff, being on X, listening to podcasts. Hire people obsessed with solving people's problems, not their own ego or glory. Hire people who add energy — the kind you'd run through a wall with. In interviews, ask how they navigate unknown, unstructured situations where it's unclear which way is forward; being energized by that is a huge credibility boost.

Lenny's Podcast · Why every company now needs to think and operate like a lab team | Josh Woodward (VP Google Labs)

Create an environment where people can say 'I don't think this is working'

As a leader, Josh Woodward recommends creating an environment where people can ask uncomfortable questions and say 'hey, I don't think this is working' — and when they do, respond with 'this is great, thank you for doing this' (he replied-all exactly that when a team member said a launch shouldn't happen). When the team says it's not going, pivot and do something different.

Lenny's Podcast · Why every company now needs to think and operate like a lab team | Josh Woodward (VP Google Labs)

Implement a retry loop that takes validation or execution errors and passes them back into the LLM for self-correction, up to a max number of attempts

Implement a retry loop that takes validation or execution errors and passes them back into the LLM for self-correction, up to a max number of attempts. The error retry loop is one of the most important components of the system. When validation fails, return an error that gets dumped out for the retry loop to consume.

AI Engineer · Beyond RAG: Build a Relational Context Engine from Scratch — Peter Werry, Unblocked

Sanitize LLM-generated queries using an allow list rather than a deny list, denying everything by default

Sanitize LLM-generated database queries using an allow list rather than a deny list, denying everything by default. Create an allow list of allowed operators and potentially an allow list of allowed fields when validating LLM query output. Allow-listing operators and fields prevents prompt injection through where or function clauses that could run JavaScript on the server.

AI Engineer · Beyond RAG: Build a Relational Context Engine from Scratch — Peter Werry, Unblocked

Checklist: set the latency budget first, then spend the remainder on model quality

Jesse Hall recommends a checklist for building voice agents: set the latency budget first, measure perceived latency, and spend whatever budget is left on model quality rather than on speed that doesn't matter.

AI Engineer · Latency Is a Budget. Humanlike Is the Goal. — Jesse Hall, LiveKit

Route agent-to-LLM traffic through a model proxy that logs, sanitizes, and strips sensitive data

Patterson recommends placing a model proxy or gateway between the agent and the LLM provider. The proxy sniffs and logs all traffic for full visibility, sanitizes content, strips prompt injections, and removes sensitive data like credit card numbers before prompts reach the provider.

AI Engineer · The Lethal Trifecta Is Already on Your Laptops — Michael Patterson, Coder

Deploy an agent firewall that blocks unapproved commands

Patterson's third architectural protection: an agentic firewall that monitors process logs and denies dangerous commands the agent hasn't been pre-approved to run, such as 'gh repo delete' or dropping a database table. Without it, an agent running unapproved commands can cause irreversible damage.

AI Engineer · The Lethal Trifecta Is Already on Your Laptops — Michael Patterson, Coder

Restrict agent external communication: VPN, VPC, firewall, and whitelisted domains

Patterson's second pillar: restrict the agent's external communication. Isolate it from your network, allow only whitelisted domains, and put the agent's VM on your VPN and behind a firewall so all agent traffic is governed and visible. He also recommends alerting admins when an agent breaches its perimeter so they can kill the agent, while noting that alert fatigue can make everything AI does become noise.

AI Engineer · The Lethal Trifecta Is Already on Your Laptops — Michael Patterson, Coder

Run AI agents in a secure remote environment, not on your laptop

Patterson's first pillar of agentic security: don't put agents on your local machine, where they can access unintended credentials and data. Instead, run them in an isolated cloud development environment accessed via URL, RDP, or SSH, with right-scoped permissions and no external communication unless explicitly allowed.

AI Engineer · The Lethal Trifecta Is Already on Your Laptops — Michael Patterson, Coder

Creating a custom agent via the CLI agent command

Use the agent command in the CLI to set up a workflow that adds an agent into your CLI environment. When creating the agent, choose whether it lives at the project level or the user level so it's available regardless of project. Copilot scaffolds the entire agent for you: description, front matter, success criteria, and methodology structure. You just write a general one-line description of what you want the agent to do and let the builder fill in the specifics.

Method, not a verified protocol.

AI Engineer · From Your Laptop to the Pipeline: Scaling Custom Agents with GitHub Copilot

Use the agentic workflows framework with Copilot CLI headless in CI for repo-resident agents

When you want an agent to run as part of the repository rather than locally, the creator points to the agentic workflows framework using the Copilot CLI in headless mode inside a CI system, which can be triggered on events happening in the repo.

AI Engineer · From Your Laptop to the Pipeline: Scaling Custom Agents with GitHub Copilot

Instrument every pipeline agent with OpenTelemetry for cost and performance analysis

The creator recommends instrumenting every agent running in the pipeline with Otel so you can go to a SIM or logging platform and analyze how agents run across repositories and how much they cost, including pivoting on average token use.

AI Engineer · From Your Laptop to the Pipeline: Scaling Custom Agents with GitHub Copilot

The app server ships in the Codex CLI binary; bundle it with your app

The app server comes as part of the Codex CLI — it's the same binary, so if you have the CLI you can spin up the app server right now. But you should bundle the app server binary with your application so you have full control over the capabilities of the version you're building against, since the protocol regularly changes between versions.

AI Engineer · Building on the Codex Harness — Dominik Kundel, OpenAI

Prefer developer instructions over writing your own system prompt

Dominik Kundel warns against rushing to write your own system prompt: OpenAI constantly tunes the harness to keep prompts in distribution and the agent performing well, and you lose all of that by replacing base instructions. If changing instructions, start by adding developer instructions rather than changing the base instructions.

AI Engineer · Building on the Codex Harness — Dominik Kundel, OpenAI

Prefer developer instructions over writing your own system prompt

Dominik Kundel advises starting with developer instructions rather than changing base instructions when customizing the agent. Rushing to write your own system prompt loses the harness team's work keeping the prompt in distribution so the agent performs well; he says to be careful if you do write your own.

AI Engineer · Building on the Codex Harness — Dominik Kundel, OpenAI

Use the app server for comprehensive clients and many parallel threads

Dominik Kundel recommends the app server when building a more comprehensive client that shows the agent, or when spinning up many Codex threads at once — such as running your own eval harness — because it gives full control over the harness and avoids hand-rolling parallelization of CodexExec calls.

AI Engineer · Building on the Codex Harness — Dominik Kundel, OpenAI

Customize the agent via developer instructions and dynamic tools

The two biggest things to look at when customizing the agent on the app server are developer instructions — appending on top of Codex's system prompt rather than replacing it, leveraging OpenAI's prompt-tuning work while customizing for your use case (this is what OpenAI does in the Codex app, since some functionality is app-specific) — and bringing tools: dynamic tools expose additional functions via MCP or by defining tools directly, and can be marked as deferred loading so Codex finds them via a tool search tool instead of polluting the context window.

AI Engineer · Building on the Codex Harness — Dominik Kundel, OpenAI

Expose new functions in your interface but mark them as deferred

Dominik Kundel recommends exposing new functions as part of your interface but marking them deferred, so you avoid polluting the context window and confusing the agent while keeping the functions available. He notes the Codex app does this so the agent can control itself, e.g. spinning up automations.

AI Engineer · Building on the Codex Harness — Dominik Kundel, OpenAI

Ship your own harness instead of relying on the user's installed CLI

Dominik Kundel advises shipping your own version of the harness to eliminate many headaches: relying on the user's installed CLI means constantly running into breaking changes (especially with experimental features, even though OpenAI tries to minimize them), and worrying about whether users have updated to get the latest features or nudging them to update.

AI Engineer · Building on the Codex Harness — Dominik Kundel, OpenAI

Ship your own version of the harness instead of relying on the user's installed CLI

Dominik Kundel recommends shipping your own harness to eliminate headaches: relying on the user's installed CLI exposes you to breaking changes (especially with experimental features) and forces you to worry about whether users updated to a version with the feature you need.

AI Engineer · Building on the Codex Harness — Dominik Kundel, OpenAI