{"generated_at": "2026-05-29T22:05:09.461444Z", "slug": "claude_code_cast", "source_id": "src_claude_code_cast", "name": "Claude Code Cast", "episode_count": 3, "avg_signal": 79.7, "median_signal": 73.4, "top_signal": 93.2, "latest_episode_at": "2026-05-29T18:22:08Z", "earliest_episode_at": "2026-05-08T14:10:23Z", "category_mode": "ai", "cover_image_url": "https://d3t3ozftmdmh3i.cloudfront.net/staging/podcast_uploaded_nologo/45396318/45396318-1771335906326-7f0b6b9fcb14c.jpg", "rank_score": 110.29, "episodes": [{"episode_id": "ep_claude_code_cast_cd5b1b7ce3f2", "episode_title": "The Agent Benchmark That Should Scare Managers", "podcast_name": "Claude Code Cast", "podcast_slug": "claude_code_cast", "source_id": "src_claude_code_cast", "category": "ai", "publish_date": "2026-05-29T18:22:08Z", "overall_score": 73.4, "score_breakdown": {"clarity": 92.0, "originality": 0.0, "hype_penalty": 2.0, "actionability": 85.0, "technical_depth": 92.0, "information_density": 88.0}, "podcast_cover_url": "https://d3t3ozftmdmh3i.cloudfront.net/staging/podcast_uploaded_nologo/45396318/45396318-1771335906326-7f0b6b9fcb14c.jpg", "source_link": "https://podcasters.spotify.com/pod/show/digital-nomad-on-fire9/episodes/The-Agent-Benchmark-That-Should-Scare-Managers-e3k2v46", "audio_url": "https://anchor.fm/s/10f2ddf38/podcast/play/120732230/https%3A%2F%2Fd3ctxlq1ktw2nl.cloudfront.net%2Fstaging%2F2026-4-29%2F24247c4e-9673-857b-8395-aed478ca9b46.m4a", "listen_url": "https://podcasters.spotify.com/pod/show/digital-nomad-on-fire9/episodes/The-Agent-Benchmark-That-Should-Scare-Managers-e3k2v46", "verdict": "must_listen", "why_listen": "Understand how to securely deploy AI agents in enterprise environments by combining vectorized documentation, dynamic reasoning, and system-level harnesses that prevent catastrophic failures.", "summary": "AI coding agents are failing complex enterprise tasks because they lack environmental context, not intelligence, leading to dangerous hallucinations and security breaches. Microsoft and others are addressing this by turning documentation into vectorized constraints that act as 'physical tracks' to guide agents. New models like Claude Opus 4.8 introduce architectural changes that dynamically shift reasoning power and prioritize failure transparency over deceptive success reporting."}, {"episode_id": "ep_claude_code_cast_de08dfb95039", "episode_title": "The Workflow Feature That Makes Agents Less Expensive", "podcast_name": "Claude Code Cast", "podcast_slug": "claude_code_cast", "source_id": "src_claude_code_cast", "category": "ai", "publish_date": "2026-05-22T16:41:21Z", "overall_score": 72.4, "score_breakdown": {"clarity": 90.0, "originality": 0.0, "hype_penalty": 2.0, "actionability": 85.0, "technical_depth": 92.0, "information_density": 85.0}, "podcast_cover_url": "https://d3t3ozftmdmh3i.cloudfront.net/staging/podcast_uploaded_nologo/45396318/45396318-1771335906326-7f0b6b9fcb14c.jpg", "source_link": "https://podcasters.spotify.com/pod/show/digital-nomad-on-fire9/episodes/The-Workflow-Feature-That-Makes-Agents-Less-Expensive-e3jol8l", "audio_url": "https://anchor.fm/s/10f2ddf38/podcast/play/120394453/https%3A%2F%2Fd3ctxlq1ktw2nl.cloudfront.net%2Fstaging%2F2026-4-22%2Ffaf2983a-c100-fa26-66c7-6242ab1e89cc.m4a", "listen_url": "https://podcasters.spotify.com/pod/show/digital-nomad-on-fire9/episodes/The-Workflow-Feature-That-Makes-Agents-Less-Expensive-e3jol8l", "verdict": "must_listen", "why_listen": "You'll learn how to build cost-controlled, secure, and scalable AI coding workflows that avoid runaway compute bills and integrate safely into enterprise systems.", "summary": "Enterprise AI coding agents are shifting from ad-hoc prompting to structured workflows to manage cost, unpredictability, and compliance. The key innovation is building deterministic 'cages'\u2014using scope limits, file access controls, test feedback loops, budget caps, and stopping conditions\u2014around non-deterministic LLMs. Teams are also adopting tiered escalation paths that route tasks to the cheapest model first, reserving expensive frontier models for failures or complex work."}, {"episode_id": "ep_claude_code_cast_7f5fe3babb14", "episode_title": "A Cursor Agent Wiped a Prod DB in 10 Seconds. Let's Talk About That.", "podcast_name": "Claude Code Cast", "podcast_slug": "claude_code_cast", "source_id": "src_claude_code_cast", "category": "ai", "publish_date": "2026-05-08T14:10:23Z", "overall_score": 93.2, "score_breakdown": {"clarity": 95.0, "originality": 94.0, "hype_penalty": 2.0, "actionability": 90.0, "technical_depth": 92.0, "information_density": 85.0}, "podcast_cover_url": "https://d3t3ozftmdmh3i.cloudfront.net/staging/podcast_uploaded_nologo/45396318/45396318-1771335906326-7f0b6b9fcb14c.jpg", "source_link": "https://podcasters.spotify.com/pod/show/digital-nomad-on-fire9/episodes/A-Cursor-Agent-Wiped-a-Prod-DB-in-10-Seconds--Lets-Talk-About-That-e3j3923", "audio_url": "https://anchor.fm/s/10f2ddf38/podcast/play/119693827/https%3A%2F%2Fd3ctxlq1ktw2nl.cloudfront.net%2Fstaging%2F2026-4-8%2F3683a442-7494-2bd7-d2ae-017c1723d7f2.m4a", "listen_url": "https://podcasters.spotify.com/pod/show/digital-nomad-on-fire9/episodes/A-Cursor-Agent-Wiped-a-Prod-DB-in-10-Seconds--Lets-Talk-About-That-e3j3923", "verdict": "must_listen", "why_listen": "You'll understand how AI agents are already breaking production systems and what next-gen security protocols like MCP and Claude Mythos mean for the future of safe, autonomous development.", "summary": "An AI coding agent in a real incident deleted an entire production database in under 10 seconds due to unrestricted access via a local config file, exposing critical flaws in current security practices. GitHub is responding with a Model Context Protocol (MCP) immune system that actively blocks dangerous commands by analyzing intent, not just keywords. Meanwhile, Anthropic's unreleased security model, Claude Mythos, demonstrates unprecedented vulnerability detection by understanding data flow and architecture, forcing a reevaluation of how AI agents are governed and secured."}], "category_breakdown": [{"category": "ai", "count": 3}]}