Explore every episode of the podcast Claude Code Cast

An expensive coding session can start with a broken cache, not a harder task. Fictional AI hosts Alex and Sam unpack Claude Code's September cache fixes, explain what the usage screen can actually tell you, and design a fair targeted-edit experiment without inventing benchmark results. Leave with a concrete way to separate wasted context from useful work.

On August 29 OpenAI ended its Cursor partnership. On September 3 ChatGPT, Claude, Grok, and Gemini were reported down almost simultaneously, and nobody has explained why. Fictional AI hosts Alex and Sam separate the two failure modes behind those headlines, cover what Fable 5.1's 75% cache price cut actually costs you in output tokens, and walk through a thirty-minute outage drill that tells you what you can still ship when your provider disappears.

Three benchmarking efforts ran an identical model through different coding-agent harnesses and reported token use varying seventy-fold. Fictional AI hosts Alex and Sam explain where harness tokens actually go, why Anthropic's Files API saves time but not money, and how to measure tokens-per-completed-task on your own repository before you switch tools.

Most coding-agent benchmarks reward contained tasks, but real repositories demand changes across boundaries, tests, migrations, and documentation. Fictional AI hosts Alex and Sam show how to run a five-part refactor trial that exposes whether an agent can preserve architecture—not merely produce a passing patch.

Passing CI can still leave code that slows down—or misleads—the next AI agent. Fictional AI hosts Alex and Sam use this week’s debate about Go and agent-friendly engineering to build a practical machine-legibility checklist, a handoff receipt, and one pro tip you can try in your next coding session.

OpenClaw’s release feed is moving faster than its labels can explain, so blind auto-update is a bad personal-automation strategy. Cleo and Dev build a Release Sentinel canary, keep telemetry local, and show how stateless MCP can shrink the trust you carry between jobs.

Claude Code’s move to a new Bun runtime is a reminder that your coding agent has a software supply chain too. Alex and Sam unpack runtime drift, model routers, reverse-engineering with agents, and a five-minute reproducibility receipt you can add to your next session.

GitHub gave its code-review agent better tools and watched cost rise while useful findings fell. Alex and Sam unpack why task-shaped instructions beat bigger toolboxes, how invisible environment details corrupt agent evals, and a five-line pro tip you can use on your next review.

This week, Alex and Sam look at why benchmark wins are a bad way to choose coding tools, what Godot's coding-agent ban reveals about mentorship, and a simple workflow for making agents show their work. If your team is still asking "which model scored highest?", this episode gives you a better test.

Small, local models are suddenly good enough for real agent chores, but the win is not replacing your smartest model. Cleo and Dev unpack lightweight extraction models, model-routing memory, browser-safe harnesses, and the OpenClaw skill that sends cheap tasks to local models while reserving expensive intelligence for the moments that matter.

AI coding agents are getting longer runs, more context, and more ways to touch production workflows, but this week made the real bottleneck obvious: authorization. Alex and Sam unpack MCP's missing enterprise auth layer, confused-deputy risks, Copilot context handling, and the verification habits that keep an agent from helping itself to the wrong keys.

Coding agents are getting better at long runs, but this week's news points at the real limit: proof. Alex and Sam unpack agent loops, Stack Overflow for Agents, Copilot CLI delegation, local-model coding, and the verification habits that keep a confident agent from shipping the wrong thing.

Coding agents are no longer just a workflow story; they are a cost, context, and control story. Alex and Sam unpack Cursor's pricing reset, Uber capping Claude Code usage, GitHub's agent-native desktop app, Microsoft Rayfin, and the spending harness every team needs before the next invoice arrives.

Agentic coding tools are moving into enterprise workflows, but the week's most useful signal is a benchmark where frontier models still struggle below 50% on real IT tasks. Alex and Sam unpack Microsoft Learn grounding, agent deception, Copilot data leaks, and the practical harness every team should build before handing agents production authority.

Claude Code workflows, enterprise Codex deployments, and rising token costs all point to the same lesson: coding agents need operating systems, not just better prompts. Alex and Sam dig into /workflows, on-prem Codex, CI for agents, and the new decision fatigue of choosing where each task should run.

OpenAI's Windows sandbox work is the practical story behind safer coding agents this week. Alex and Sam dig into Codex on Windows, remote cloud coding agents, Claude Code billing splits, and why a Raspberry Pi running rm -rf is the warning label every agent workflow needs.

A Cursor AI agent deleted PocketOS's entire production database on April 25th — in under 10 seconds. This week Alex and Sam dig into the AI agent credential crisis, Anthropic's wild SpaceX/xAI compute deal, Mozilla using Claude to find hundreds of Firefox vulnerabilities, and whether OpenAI Codex is actually closing the gap on Claude Code. If you've ever given an agent database access, listen before your next deploy.

Anthropic's Claude Security tool just dropped out of closed preview and it will scan your entire codebase for vulnerabilities — and the results might be uncomfortable. This week we also dig into Cursor's $60 billion bet on being the "harness" rather than the model, why AI agents are literally forcing developers to keep their laptops open, and the Zig project's nuclear take on AI contributions. If you write code with AI help, this episode is required listening.

Turns out the Claude Code quality complaints weren't in your head — three separate bugs in the harness quietly degraded your results for two months, and Anthropic just confirmed it. This week: the $100/month pricing scare that wasn't, Claude Mythos fixing 271 Firefox vulnerabilities, the SpaceX-Cursor deal that changes the competitive landscape, and why the Claude Code creator says your cloud-native workflow is probably wrong. Essential listening before your next session.

Claude Opus 4.7 is here with upgraded vision, memory, and instruction-following — but Simon Willison's pelican benchmark just handed the win to a local Alibaba model running on a laptop. We dig into what that actually means, plus Anthropic's new identity verification layer, Amazon's MCP bet, and whether "personal software" is about to change who gets to be a developer. Your commute just got more interesting.

A Reddit user just proved that Claude Code's "max effort" thinking mode has been silently failing since v2.0.64 — and most of us never noticed. This week: the bug, the fix, and what it says about trusting your tools. Plus, Anthropic launches Claude Managed Agents, OpenAI goes to $100/month to poach Claude Code users, and the AI-generated PR crisis that's about to hit enterprise teams hard. Required listening before you open your terminal Monday morning.

Claude Code's source code leaked — accidentally — and the internet went digging. This week Alex and Sam tear through what the leak actually revealed, why it matters for how you use Claude Code today, and why your CI/CD pipeline is quietly becoming the new bottleneck. Plus: GitHub Copilot just shipped parallel agents and the usage limit complaints are getting loud. Don't skip this one.

GitHub Copilot literally edited an advertisement into a developer's pull request this week — and that's somehow not even the most alarming Copilot story. We dig into GitHub's new policy to train on your code, the cache bugs silently inflating Claude Code API bills by 10-20x, and Boris Cherny's 15 hidden Claude Code features. This one's got receipts.

A senior engineer's post about merging vibe-coded PRs full of "confident spaghetti" is going viral — code that looks clean, passes all tests, and hides a race condition nobody understands. This week Alex and Sam dig into whether the backlash is fair, what it actually exposes about code review culture, and why the ratio of writing-to-understanding cost just inverted. Plus: Claude Code can now spawn parallel subagents natively, MCP servers just crossed 800 and counting, and a 7-year-old production bug got fixed in 90 minutes.

A developer lost 25,000 documents in a single AI agent session — and wrote up exactly how it happened. This week Alex and Sam dig into the horror story, what safeguards actually work, and why "I'll just be careful" isn't a strategy. Plus: Shopify's CEO is submitting his own performance PRs using Claude, Simon Willison's case that AI *should* raise code quality (not tank it), and the Claude Code gateway daemon you probably want running right now.

This week: Anthropic announces Auto Mode to kill permission fatigue, GPT-5.4 drops in Cursor, and a 30-year veteran explains why he hasn't written code in six months. Plus: why your AI-generated tests are lying to you, and our battle-tested prompt for sending Claude on a bug-hunting crusade. Essential listening for anyone shipping with AI.

Claude Code's new auto-memory feature changes the game for long-term projects, France deploys a government-wide MCP server, and we dig into why shipping 118 commits/day might not be the flex you think it is. Plus: the Lovable security disaster that proves AI-generated code needs human eyes.

Anthropic just banned the use of consumer OAuth tokens in third-party tools like Cline and Roo Code — and sent OpenClaw a cease and desist (OpenAI promptly hired the dev). Meanwhile, Cursor 2.5 launches a full plugin marketplace with partners like Stripe, Figma, and AWS, plus async background agents and sandbox access controls.

We also cover a serious data isolation bug in Claude for PowerPoint that served one user another person's legal documents. Pro Tip: the MCP server that claims to cut refactoring costs by 95% by giving AI surgical precision instead of a sledgehammer.

One-prompt app debate, Auto Memory, Claude Code skills replacing startups, the Pentagon-Anthropic controversy, and a community project spotlight.