From 1a9689af1d78e83dbcbf0bbb08b615cd41ed2e2c Mon Sep 17 00:00:00 2001
From: Emeriko <dietrich.gebert@gmail.com>
Date: Thu, 18 Jun 2026 16:48:20 +0200
Subject: [PATCH] docs(readme): add agentic benchmark chart (LOC per arm)

Bar chart of mean LOC per task by arm (Haiku 4.5) in the Numbers section, same
system-gray style as the single-shot chart so it reads on both GitHub themes.
The old single-shot chart stays in the collapsed "isolated generation" section.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
---
 README.md                    |  4 ++++
 assets/benchmark-agentic.svg | 23 +++++++++++++++++++++++
 2 files changed, 27 insertions(+)
 create mode 100644 assets/benchmark-agentic.svg

diff --git a/README.md b/README.md
index 34c97a3..238c1be 100644
--- a/README.md
+++ b/README.md
@@ -46,6 +46,10 @@ More survivors in [examples/](examples/).
 
 The honest measurement is a real agent doing real work: a headless Claude Code session editing [tiangolo's full-stack-fastapi-template](https://github.com/fastapi/full-stack-fastapi-template) (a real FastAPI + React repo), scored on the `git diff` it leaves behind. Twelve feature tickets, the same agent with and without the skill, n=4, Haiku 4.5.
 
+<p align="center">
+  <img src="assets/benchmark-agentic.svg" width="860" alt="Mean lines of code per task by arm on the agentic benchmark (Haiku 4.5): baseline 191, caveman 152, yagni-oneliner 128, ponytail 87">
+</p>
+
 | vs no-skill baseline | LOC | tokens | cost | time | safe |
 |---|--:|--:|--:|--:|--:|
 | **ponytail** | **-54%** | **-22%** | **-20%** | **-27%** | **100%** |
diff --git a/assets/benchmark-agentic.svg b/assets/benchmark-agentic.svg
new file mode 100644
index 0000000..6903706
--- /dev/null
+++ b/assets/benchmark-agentic.svg
@@ -0,0 +1,23 @@
+<svg viewBox="0 0 860 268" xmlns="http://www.w3.org/2000/svg" font-family="-apple-system, 'Segoe UI', Helvetica, Arial, sans-serif">
+  <title>Mean lines of code per task by arm, agentic benchmark on Haiku 4.5</title>
+  <text x="20" y="26" font-size="15" font-weight="600" fill="#8b949e">Lines of code per task (12 real-repo features), Claude Code on Haiku 4.5</text>
+  <text x="20" y="46" font-size="12" fill="#8b949e" opacity="0.85">Mean over 12 frontend + backend tickets, n=4. Lower is leaner. ponytail writes 54% less than a no-skill agent and stays 100% safe.</text>
+
+  <text x="132" y="91" font-size="13" fill="#8b949e" text-anchor="end">baseline</text>
+  <rect x="140" y="78" width="520" height="18" rx="3" fill="#8b949e"/>
+  <text x="668" y="91" font-size="12" fill="#8b949e">191</text>
+
+  <text x="132" y="129" font-size="13" fill="#8b949e" text-anchor="end">caveman</text>
+  <rect x="140" y="116" width="413" height="18" rx="3" fill="#d9822b"/>
+  <text x="561" y="129" font-size="12" fill="#d9822b">152 (-20%)</text>
+
+  <text x="132" y="167" font-size="13" fill="#8b949e" text-anchor="end">yagni-oneliner</text>
+  <rect x="140" y="154" width="348" height="18" rx="3" fill="#8957e5"/>
+  <text x="496" y="167" font-size="12" fill="#8957e5">128 (-33%)</text>
+
+  <text x="132" y="205" font-size="13" font-weight="600" fill="#2da44e" text-anchor="end">ponytail</text>
+  <rect x="140" y="192" width="237" height="18" rx="3" fill="#2da44e"/>
+  <text x="385" y="205" font-size="12" font-weight="600" fill="#2da44e">87 (-54%)</text>
+
+  <text x="20" y="248" font-size="11" fill="#8b949e" opacity="0.8">Range by task: -94% on over-build features (date/color picker, native input vs custom component) to ~0% on already-minimal backend code. LOC = git diff added lines. Writeup: benchmarks/results/2026-06-18-agentic.md</text>
+</svg>