Agentic benchmark answering #126 (real repo, LOC + safety)#158
Merged
Conversation
Rebuild the benchmark to the standard #126 asked for: real headless Claude Code sessions (not a bare model) editing a real public repo (tiangolo/full-stack-fastapi-template @ cd83fc1, MIT), fair arms (baseline, caveman, ponytail, and the "YAGNI + one-liners" prompt), n=4, Haiku 4.5. LOC is the git diff; the safety tasks execute the produced code against adversarial input. Results: ponytail -54% LOC mean (up to -94% on over-build features like the date/color picker), -22% tokens, -20% cost, -27% time, and never more than baseline; 100% safe vs the one-liner prompt's 95% (it dropped a path-traversal guard once). caveman writes less code but spends more tokens. Also fixes a baseline-contamination bug (the ponytail plugin's SessionStart hook fired on every arm; now isolated with --setting-sources project,local + per-arm --plugin-dir) and a Windows subprocess-timeout hang. Lead both READMEs with the agentic numbers; demote the single-shot 80-94% to a labelled "isolated generation" note; supersede the contaminated 2026-06-17 writeup. Dead react-app fixture left untracked. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
DietrichGebert
added a commit
that referenced
this pull request
Jun 18, 2026
Grouped bars of LOC, tokens, cost and time as a % of the no-skill baseline (lower is leaner/cheaper/faster), plus a separate safety strip (baseline, caveman and ponytail 100%; yagni-oneliner 95%). System-gray palette so it reads on both GitHub themes. The chart commits landed after #158 had already squash-merged, so this brings the chart onto main. Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Rebuilds the ponytail benchmark to the standard #126 asked for.
What
Real headless Claude Code sessions (not a bare model) editing a real public repo (
tiangolo/full-stack-fastapi-template@cd83fc1, MIT). Fair arms: baseline, caveman, ponytail, and the "YAGNI + one-liners" prompt from the #126 writeup. n=4, Haiku 4.5.git diffon 12 real feature tickets.Results (vs no-skill baseline)
ponytail cuts the most code (up to -94% on over-build features like the date/color picker, where it reaches for a native
<input>instead of a component), is a wash on irreducible code, never writes more, and is the only arm that stays 100% safe. The one-liner prompt dropped a path-traversal guard once.Also in this PR
SessionStarthook fired on every arm; isolated now with--setting-sources project,local+ per-arm--plugin-dir) and a Windows subprocess-timeout hang.Writeup:
benchmarks/results/2026-06-18-agentic.md· reproduction:benchmarks/agentic/README.md.Addresses #126.
🤖 Generated with Claude Code