Context-Engineering on ROBOCO

Surprisingly, the Ralph Loop Is 9x Cheaper - The Economics of Context Caching

Sun, 26 Jul 2026 10:00:00 +0900

A single-session loop that just states a goal and runs until done finished the same task at one-ninth the cost of a plan-driven workflow. The secret isn’t the loop — it’s the cache.

Dohyun Jung - Principal Consultant, ROBOCO

The Ralph loop is a simple approach. You specify only the goal and the completion criteria, then let the agent iterate autonomously in a single session until it passes the completion check. No plan document, no task decomposition, no handoff notes between sessions. At first glance it looks unplanned and inefficient.

Does Karpathy's LLM Wiki Actually Work? A 72-Run Benchmark

Thu, 07 May 2026 11:00:00 +0900

“To keep agents from re-deriving everything every time, you need a compiled knowledge artifact.” — Andrej Karpathy, LLM Wiki (GitHub Gist, 2026-04)¹

Dohyun Jung - Principal Consultant, ROBOCO

TL;DR

We measured whether Andrej Karpathy’s LLM Wiki pattern (“don’t re-read every time; reuse knowledge compiled once”¹) is actually a win, using an 8-task × 3-technique × 3-repetition = 72-run benchmark in a 30-repository migration workspace.
LLM Wiki placed first on all three dimensions: tokens, time, and quality. Versus Vanilla, 54% fewer tokens and 39% less time (statistically a large effect, §4), with quality equal to Vanilla and better than Graphify.
Graphify (GraphRAG) is unsuitable as a default tool. Negligible token savings, the longest time, and the lowest quality.
That said, it’s not universal. Wiki wins decisively on tasks that synthesize multiple sources, but for tasks whose answer sits plainly in a single source (exhaustive grep, looking up a specific table), reading directly is faster and more accurate (§3).
On receiving these results, the project’s default context-retrieval tool was switched to LLM Wiki that same evening.

Technique	Tokens (avg)	Time (avg)	Quality (out of 25)
Vanilla (direct read/grep)	755K	167s	15.1
LLM Wiki	350K	101s	16.0
Graphify (GraphRAG)	617K	180s	13.6

Below is the experimental design and detailed numbers that support this conclusion.

A Token Management Strategy for Vibe Coding

Thu, 19 Mar 2026 09:00:00 +0900

Running out of tokens is rarely a model performance problem. It is usually a problem with how you operate your context.

Dohyun Jung - Principal Consultant, ROBOCO

Use a vibe coding tool like Claude Code, Codex, or Gemini long enough and the same symptoms start showing up. Responses slow down, constraints you already agreed on get forgotten, and the model starts touching files that have nothing to do with the task. We usually describe this as “running out of tokens,” but what is actually happening is closer to context pollution, or context rot.