Michael

## Summary - expose per-plugin `numTests` and `fullDataset` options in the redteam UI config - support these options in dataset plugins ## Testing - `npx eslint "src/app/src/pages/redteam/setup/types.ts" "src/app/src/pages/redteam/setup/hooks/useRedTeamConfig.ts" "src/app/src/pages/redteam/setup/components/Plugins.tsx" "src/app/src/pages/redteam/setup/components/PluginConfigDialog.tsx"...

## Summary - add Aegis dataset plugin and grader - register Aegis plugin in constants and plugin index - document Aegis plugin and expose in docs - update config schema...

## Summary - implement Tau strategy using SimulatedUser provider - expose redteam provider `promptfoo:redteam:tau` - document Tau in strategy docs and presets - support Tau in redteam setup UI -...

feat(cli): track per provider token usage statistics

1

This PR introduces tracking of token usage statistics per provider. - Adds detailed token and character usage tracking. - Implements CLI table display for usage summary.

Enables SQLite WAL mode to reduce locking. Need to test before merging.

refactor: avoid process exit

2

## Summary - replace direct `process.exit()` calls with `process.exitCode` - bail early from functions instead of exiting the process - update tests for the new exit behavior ## Testing -...

chore(dependencies): update GitHub Actions and npm dependencies

2

- Updated GitHub Actions workflows to use newer versions of actions. - Upgraded npm dependencies to their latest versions. - Updated Node.js version requirements in documentation and configuration files. No...

feat(webui): add user-rated filter to show only manually rated results

8

## Summary Adds a new filter mode "Show user-rated only" to the eval results view, allowing users to filter and display only results that have been manually rated using the...

feat: add OpenAI Guardrails redteam plugin

2

## Summary Add new dataset-based redteam plugin to test jailbreak resistance using OpenAI's official guardrails evaluation dataset. ## Key Features - **Dataset Source**: Uses OpenAI's [guardrails-python evaluation demo](https://github.com/openai/openai-guardrails-python/tree/main/src/guardrails/evals/eval_demo) dataset -...