Skip to content
Merged
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension


Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
8 changes: 4 additions & 4 deletions .github/workflows/evaluations.yaml
Original file line number Diff line number Diff line change
Expand Up @@ -10,14 +10,14 @@ on:
- 'master'
# Also run on PRs with 'evals' label for testing
pull_request:
types: [labeled, synchronize, reopened]
types: [labeled, synchronize, reopened, closed]

jobs:
evaluations:
name: MCP tool calling evaluations
runs-on: ubuntu-latest
# Run on master pushes or PRs with 'evals' label
if: github.event_name == 'push' || contains(github.event.pull_request.labels.*.name, 'validated')
# Run on master pushes, merged PRs, or PRs with 'validated' label
if: github.event_name == 'push' || (github.event_name == 'pull_request' && github.event.pull_request.merged == true) || contains(github.event.pull_request.labels.*.name, 'validated')

steps:
- name: Checkout code
Expand All @@ -39,7 +39,7 @@ jobs:
- name: Run evaluations
run: npm run evals:run
env:
GITHUB_PR_NUMBER: ${{ github.event_name == 'pull_request' && github.event.number || 'master' }}
GITHUB_PR_NUMBER: ${{ github.event_name == 'pull_request' && github.event.number || (github.event_name == 'push' && 'master') || 'unknown' }}
PHOENIX_API_KEY: ${{ secrets.PHOENIX_API_KEY }}
PHOENIX_BASE_URL: ${{ secrets.PHOENIX_BASE_URL }}
OPENROUTER_BASE_URL: ${{ secrets.OPENROUTER_BASE_URL }}
Expand Down