Skip to content

Feature/skills aggregator: add SkillEvalAggregator for batch evaluation comparisons - #229

Draft
venkatkrish543re wants to merge 9 commits into
strands-agents:mainfrom
venkatkrish543re:feature/skills-aggregator
Draft

venkatkrish543re wants to merge 9 commits into
strands-agents:mainfrom
venkatkrish543re:feature/skills-aggregator

Conversation

@venkatkrish543re

@venkatkrish543re venkatkrish543re commented May 14, 2026 •

Copy link
Copy Markdown
Contributor

No description provided.

ybdarrenwang and others added 9 commits May 6, 2026 19:21
Adds skills/ subpackage providing paired-comparison aggregation for evaluating agent skills against a baseline. Mirrors the chaos aggregator pattern from feature/aggregator-demo-2.

- SkillEvalAggregator with Wilcoxon, paired-t, and McNemar tests

- Bootstrap CI on the mean delta (1000 resamples)

- Corruption filtering before paired statistics

- SkillEvalExperiment composes base Experiment

- Rich-based interactive display

- 44 unit tests covering paired stats, corruption filtering, pairing, and serialization

Closes strands-agents#228
@yonib05 yonib05 added area-evaluators Evaluators: output, trajectory, tool use, interactions, and LLM-as-judge quality metrics enhancement New feature or request labels Jun 11, 2026

This branch had an error being deployed

1 failed deployment
manual-approval — b8dfeeec Deployed May 14, 2026 by venkatkrish543re via Trigger Strands Review #5
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

area-evaluators Evaluators: output, trajectory, tool use, interactions, and LLM-as-judge quality metrics enhancement New feature or request

Projects

None yet

Development

Successfully merging this pull request may close these issues.

3 participants