loc.js regex was missing \r? before \n, so CRLF code fences were
never matched and the metric fell back to counting the whole response.
Also removed dead codeOf() from behavior.js (defined but never called).
Fixes#339
loc.js filtered comments line by line, so /* ... */ block comments whose
continuation lines are not *-aligned had their inner lines counted as code.
A plain indented block comment scored higher than the same code written
JSDoc-style. Strip /* ... */ before the line count so both are equal, and
add loc.test.js to lock the behavior.
Fixes#231
Fixes the local benchmark LOC counter (counted only fenced code, scored bare output 0), makes summary output ASCII-safe (a Unicode arrow crashed the script on Windows cp1252), gitignores generated artifacts, and refreshes the llama3.2 writeup with n=5 data showing the LOC effect is within the noise floor. Follow-up to #63. Verified live.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Commit a promptfoo harness (config + arm prompts + LOC metric + vendored
caveman SKILL) so anyone can re-run the comparison: no-skill vs caveman vs
ponytail, across Haiku / Sonnet / Opus, 10 runs per cell, median reported.
Replace the old unreproducible 6-task chart with assets/benchmark-3model.svg
from this run, and reframe the README to the reproducible numbers: ponytail
writes 80-94% less code, costs 47-77% less, and runs 3-6x faster than a
no-skill agent on every model. benchmarks/README.md carries the median tables
and the reproduce command. Drops nothing that is not measured.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Aly Dhedhi