Dataset Viewer
Auto-converted to Parquet Duplicate
namespace
stringclasses
1 value
skill
stringclasses
1 value
case
stringlengths
1
3
model
stringclasses
2 values
passed
bool
2 classes
task
stringclasses
6 values
latency_ms
float64
3.32k
37.8k
input_tokens
int64
3
1.85k
output_tokens
int64
224
2.56k
cost_usd
float64
0
0.05
timestamp
timestamp[s]date
2026-05-19 20:30:41
2026-05-19 23:53:12
drupal.org/ai_best_practices
drupal-automated-testing
B01
anthropic/claude-haiku-4-5-20251001
true
A complete PHP Kernel test file using KernelTestBase with correct namespace, RunTestsInSeparateProcesses attribute, and explicit setUp with installConfig/installSchema.
10,000
3
638
0.018185
2026-05-19T21:08:31
drupal.org/ai_best_practices
drupal-automated-testing
B02
anthropic/claude-haiku-4-5-20251001
true
A complete PHP Functional test file using BrowserTestBase with correct namespace and RunTestsInSeparateProcesses attribute.
11,400
3
681
0.018835
2026-05-19T21:08:42
drupal.org/ai_best_practices
drupal-automated-testing
B03
anthropic/claude-haiku-4-5-20251001
true
A complete FunctionalJavascript test using WebDriverTestBase with waitForElement wrapped in assertNotEmpty to avoid flaky assertions.
37,800
3
2,558
0.047014
2026-05-19T21:09:20
drupal.org/ai_best_practices
drupal-automated-testing
B04
anthropic/claude-haiku-4-5-20251001
true
A corrected setUp method showing installEntitySchema and installConfig calls that create the required database tables.
10,000
3
423
0.014928
2026-05-19T21:09:30
drupal.org/ai_best_practices
drupal-automated-testing
B05
anthropic/claude-haiku-4-5-20251001
true
A minimal Kernel test skeleton with KernelTestBase, correct namespace, and RunTestsInSeparateProcesses attribute.
7,700
3
417
0.014821
2026-05-19T21:09:38
drupal.org/ai_best_practices
drupal-automated-testing
B06
anthropic/claude-haiku-4-5-20251001
false
Uses the \Drupal\Tests\my_commerce\FunctionalJavascript namespace and extends WebDriverTestBase.
6,300
3
386
0.014395
2026-05-19T21:09:44
drupal.org/ai_best_practices
drupal-automated-testing
1
anthropic/claude-haiku-4-5-20251001
true
A complete PHP Kernel test file using KernelTestBase with correct namespace, RunTestsInSeparateProcesses attribute, and explicit setUp with installConfig/installSchema.
6,840.9
1,844
941
0.005239
2026-05-19T23:53:12
drupal.org/ai_best_practices
drupal-automated-testing
2
anthropic/claude-haiku-4-5-20251001
true
A complete PHP Functional test file using BrowserTestBase with correct namespace and RunTestsInSeparateProcesses attribute.
7,354.1
1,852
940
0.005242
2026-05-19T23:53:12
drupal.org/ai_best_practices
drupal-automated-testing
3
anthropic/claude-haiku-4-5-20251001
true
A complete FunctionalJavascript test using WebDriverTestBase with waitForElement wrapped in assertNotEmpty to avoid flaky assertions.
12,429
1,847
1,819
0.008754
2026-05-19T23:53:12
drupal.org/ai_best_practices
drupal-automated-testing
4
anthropic/claude-haiku-4-5-20251001
true
A corrected setUp method showing installEntitySchema and installConfig calls that create the required database tables.
5,031.9
1,831
431
0.003189
2026-05-19T23:53:12
drupal.org/ai_best_practices
drupal-automated-testing
5
anthropic/claude-haiku-4-5-20251001
true
A minimal Kernel test skeleton with KernelTestBase, correct namespace, and RunTestsInSeparateProcesses attribute.
3,321.2
1,818
337
0.002802
2026-05-19T23:53:12
drupal.org/ai_best_practices
drupal-automated-testing
6
anthropic/claude-haiku-4-5-20251001
true
Uses the \Drupal\Tests\my_commerce\FunctionalJavascript namespace and extends WebDriverTestBase.
4,233.4
1,844
446
0.003259
2026-05-19T23:53:12
drupal.org/ai_best_practices
drupal-automated-testing
B01
mistralai/mistral-large-2512
true
A complete PHP Kernel test file using KernelTestBase with correct namespace, RunTestsInSeparateProcesses attribute, and explicit setUp with installConfig/installSchema.
13,800
1,614
850
0.002082
2026-05-19T20:30:41
drupal.org/ai_best_practices
drupal-automated-testing
B02
mistralai/mistral-large-2512
true
A complete PHP Functional test file using BrowserTestBase with correct namespace and RunTestsInSeparateProcesses attribute.
9,300
1,621
623
0.001745
2026-05-19T20:30:51
drupal.org/ai_best_practices
drupal-automated-testing
B03
mistralai/mistral-large-2512
true
A complete FunctionalJavascript test using WebDriverTestBase with waitForElement wrapped in assertNotEmpty to avoid flaky assertions.
13,800
1,615
812
0.002025
2026-05-19T20:31:05
drupal.org/ai_best_practices
drupal-automated-testing
B04
mistralai/mistral-large-2512
true
A corrected setUp method showing installEntitySchema and installConfig calls that create the required database tables.
4,400
1,601
266
0.0012
2026-05-19T20:31:09
drupal.org/ai_best_practices
drupal-automated-testing
B06
mistralai/mistral-large-2512
true
Uses the \Drupal\Tests\my_commerce\FunctionalJavascript namespace and extends WebDriverTestBase.
3,700
1,614
224
0.001143
2026-05-19T20:31:13

Drupal AI Eval Results (POC)

Evaluation results from the ai_best_practices skill suite, stored in EvalEval envelope format (schema version 0.2.2 / instance_level_eval_0.2.2).

The Dataset Viewer above shows a flat summary of all results. The full envelope files (with prompts, model responses, and assertion details) live under data/.

Structure

results.jsonl                  # flat summary (one row per case, for the viewer)
data/
  {skill-name}/
    {developer}/
      {model-name}/
        {uuid}.json            # aggregate run record
        {uuid}_samples.jsonl   # per-instance records (one line each)

Generating

# Run evals and save traces
python3 evals/run-evals.py --behavioral \
    --skill drupal-automated-testing \
    --provider claude \
    --output-dir ./traces

# Convert traces to EvalEval envelope
python3 evals/convert_to_envelope.py \
    --trace-dir ./traces/drupal-automated-testing \
    --output-dir ./envelope \
    --skill drupal-automated-testing \
    --validate

# Upload (also writes results.jsonl for the viewer)
python3 evals/upload_to_hf.py \
    --envelope-dir ./envelope \
    --repo webchick/eval-results-poc
Downloads last month
100

Space using webchick/eval-results-poc 1