The 90% Problem: Why Most AI Agents Are Still Broken
Building an AI agent that works is easy. Building one that doesn't break is 90% of the work. Here's what that 90% actually looks like — from leaked source code and production A/B data.
Read More[ Blog_Posts ]
Claude Code + Codex Plugin: Two AI Brains, One Terminal
How to use OpenAI's Codex plugin inside Claude Code — turning Claude Opus and GPT-5.4 into a dual-brain coding system. Setup, commands, rescue workflows, and when each brain wins.
Read More
Claude Code Deep Dive Part 4: Why It Uses Markdown Files Instead of Vector DBs
Claude Code's memory system looks simple on purpose. This piece breaks down the tradeoffs behind Markdown memories, Sonnet side-queries, and the decision to avoid vector databases.
Read More
Claude Code Context Compression: the 5-Level Pipeline
How Claude Code keeps 1M-token sessions alive: 5 levels of context compression, the dual-path algorithm, and what each level drops first.
Read More
Claude Code Deep Dive Part 2: The 1,421-Line While Loop That Runs Everything
Inside query.ts — the 1,729-line async generator that is Claude Code's beating heart. 10 steps per iteration, 9 continue points, 4-stage compression, and streaming tool execution. With line numbers.
Read More
Observability and Billing for AI API Calls: A T-Shaped Architecture
AI API calls are unlike ordinary RPC: per-request cost varies 100×, tokens and models are first-class, streaming muddies timing, caching changes the pricing. A T-shaped instrumentation architecture — shared stem, specialized arms — that handles tracing, billing, and cost analytics without any of them contaminating the others.
Read More
Claude Code Frontmatter: the 4 MEMORY.md Types Explained
The 4 frontmatter types in Claude Code's MEMORY.md (user, feedback, project, reference): what each is for, when the LLM picks it, and where it breaks at scale.
Read More
Claude Code Source Leaked: Kairos, Undercover Mode, Ultraplan — 5 Hidden Features (510K Lines)
Claude Code v2.1.88 accidentally exposed 510K lines. The 5 hidden features: Kairos (permanent memory), Undercover Mode (stealth), Ultraplan (deep planning), Pet System (Buddy), Multi-Agent. Source-cited.
Read More
Consistency in Distributed Systems: Scenarios, Trade-offs, and What Actually Works
Strong, eventual, causal, read-your-writes, linearizable — consistency models are taught as a taxonomy. Production uses them as a menu. Ten scenarios, the right consistency choice for each, and the engineering that makes the choice work.
Read More
The AI Stack Explained: LLM Talks, Program Walks
A first-principles breakdown of the entire AI stack — from LLM to Agent in one mental model. An LLM can only output text. Everything else is the program.
Read More