Merge branch 'main' of https://github.com/microsoft/BC-Bench #36

	name: CI

	on:
	push:
	branches: [main]
	pull_request:
	branches: [main]
	workflow_dispatch:

	permissions:
	contents: read

	env:
	EVALUATION_RESULTS_DIR: evaluation_results

	jobs:
	lint-and-test:
	runs-on: ubuntu-latest

	steps:
	- name: Checkout repository
	uses: actions/checkout@v4

	- name: Setup Python with UV
	uses: ./.github/actions/setup-python-uv
	with:
	all-extras: 'true'

	- name: Run Ruff
	run: uv run ruff check --output-format=github .

	- name: Run tests with coverage
	run: uv run pytest --cov=src/bcbench --cov-report=term-missing

	mock-evaluation:
	runs-on: ubuntu-latest
	outputs:
	results-dir: ${{ env.EVALUATION_RESULTS_DIR }}
	strategy:
	fail-fast: false
	matrix:
	entry:
	- microsoftInternal__NAV-210528
	- microsoftInternal__NAV-224009
	- microsoftInternal__NAV-211710
	name: ${{ matrix.entry }}
	steps:
	- name: Checkout repository
	uses: actions/checkout@v4

	- name: Setup Python with UV
	uses: ./.github/actions/setup-python-uv

	- name: Run mock evaluation for ${{ matrix.entry }}
	run: \|
	uv run bcbench evaluate mock "${{ matrix.entry }}" --output-dir evaluation_results --run-id ${{ github.run_id }}
	shell: pwsh

	- name: Upload mock evaluation results
	uses: actions/upload-artifact@v4
	with:
	name: ${{ matrix.entry }}
	path: ${{ env.EVALUATION_RESULTS_DIR }}/
	retention-days: 1

	summarize-results:
	needs: mock-evaluation
	if: always()
	uses: ./.github/workflows/summarize-results.yml
	with:
	results-dir: ${{ needs.mock-evaluation.outputs.results-dir }}
	model: 'mock-model'
	agent: 'mock-agent'
	mock: true
	secrets: inherit

Provide feedback