Agent Evals (Nightly) #48
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| name: Agent Evals (Nightly) | |
| on: | |
| schedule: | |
| # Run at 3:00 AM UTC every day | |
| - cron: '0 3 * * *' | |
| workflow_dispatch: | |
| inputs: | |
| providers: | |
| description: 'Providers to run (comma-separated: mock,gemini,ollama)' | |
| required: false | |
| default: 'mock,gemini' | |
| suite: | |
| description: 'Eval suite path' | |
| required: false | |
| default: 'scripts/evals/agent-smoke.jsonl' | |
| env: | |
| NODE_VERSION: '20' | |
| PNPM_VERSION: '10' | |
| jobs: | |
| eval-mock: | |
| name: Eval (Mock) | |
| runs-on: ubuntu-latest | |
| steps: | |
| - name: Checkout | |
| uses: actions/checkout@v4 | |
| - name: Setup pnpm | |
| uses: pnpm/action-setup@v4 | |
| with: | |
| version: ${{ env.PNPM_VERSION }} | |
| - name: Setup Node.js | |
| uses: actions/setup-node@v4 | |
| with: | |
| node-version: ${{ env.NODE_VERSION }} | |
| cache: 'pnpm' | |
| - name: Install dependencies | |
| run: pnpm install --frozen-lockfile | |
| - name: Run mock evals | |
| run: | | |
| pnpm agent:eval \ | |
| --provider mock \ | |
| --suite ${{ github.event.inputs.suite || 'scripts/evals/agent-smoke.jsonl' }} \ | |
| --vault src/data/demo-files \ | |
| --json eval-results-mock.json | |
| - name: Upload mock results | |
| uses: actions/upload-artifact@v4 | |
| with: | |
| name: eval-results-mock | |
| path: eval-results-mock.json | |
| retention-days: 30 | |
| eval-gemini: | |
| name: Eval (Gemini) | |
| runs-on: ubuntu-latest | |
| if: ${{ github.event_name == 'schedule' || contains(github.event.inputs.providers, 'gemini') }} | |
| steps: | |
| - name: Checkout | |
| uses: actions/checkout@v4 | |
| - name: Setup pnpm | |
| uses: pnpm/action-setup@v4 | |
| with: | |
| version: ${{ env.PNPM_VERSION }} | |
| - name: Setup Node.js | |
| uses: actions/setup-node@v4 | |
| with: | |
| node-version: ${{ env.NODE_VERSION }} | |
| cache: 'pnpm' | |
| - name: Install dependencies | |
| run: pnpm install --frozen-lockfile | |
| - name: Run Gemini evals | |
| env: | |
| GEMINI_API_KEY: ${{ secrets.GEMINI_API_KEY }} | |
| run: | | |
| pnpm agent:eval \ | |
| --provider gemini \ | |
| --model gemini-2.0-flash \ | |
| --suite ${{ github.event.inputs.suite || 'scripts/evals/agent-smoke.jsonl' }} \ | |
| --vault src/data/demo-files \ | |
| --json eval-results-gemini.json \ | |
| --debug | |
| - name: Upload Gemini results | |
| uses: actions/upload-artifact@v4 | |
| with: | |
| name: eval-results-gemini | |
| path: eval-results-gemini.json | |
| retention-days: 30 | |
| eval-ollama: | |
| name: Eval (Ollama) | |
| runs-on: ubuntu-latest | |
| if: ${{ contains(github.event.inputs.providers || '', 'ollama') }} | |
| steps: | |
| - name: Checkout | |
| uses: actions/checkout@v4 | |
| - name: Setup pnpm | |
| uses: pnpm/action-setup@v4 | |
| with: | |
| version: ${{ env.PNPM_VERSION }} | |
| - name: Setup Node.js | |
| uses: actions/setup-node@v4 | |
| with: | |
| node-version: ${{ env.NODE_VERSION }} | |
| cache: 'pnpm' | |
| - name: Install dependencies | |
| run: pnpm install --frozen-lockfile | |
| - name: Install Ollama | |
| run: | | |
| curl -fsSL https://ollama.com/install.sh | sh | |
| ollama serve & | |
| sleep 5 | |
| - name: Pull model | |
| run: ollama pull llama3-groq-tool-use | |
| - name: Run Ollama evals | |
| run: | | |
| pnpm agent:eval \ | |
| --provider ollama \ | |
| --model llama3-groq-tool-use \ | |
| --suite ${{ github.event.inputs.suite || 'scripts/evals/agent-smoke.jsonl' }} \ | |
| --vault src/data/demo-files \ | |
| --json eval-results-ollama.json \ | |
| --debug | |
| - name: Upload Ollama results | |
| uses: actions/upload-artifact@v4 | |
| with: | |
| name: eval-results-ollama | |
| path: eval-results-ollama.json | |
| retention-days: 30 | |
| summary: | |
| name: Summary | |
| runs-on: ubuntu-latest | |
| needs: [eval-mock, eval-gemini] | |
| if: always() | |
| steps: | |
| - name: Download all artifacts | |
| uses: actions/download-artifact@v4 | |
| with: | |
| path: results | |
| pattern: eval-results-* | |
| merge-multiple: false | |
| - name: Generate summary | |
| run: | | |
| echo "## Agent Eval Results" >> $GITHUB_STEP_SUMMARY | |
| echo "" >> $GITHUB_STEP_SUMMARY | |
| echo "| Provider | Passed | Failed | Total |" >> $GITHUB_STEP_SUMMARY | |
| echo "|----------|--------|--------|-------|" >> $GITHUB_STEP_SUMMARY | |
| for dir in results/eval-results-*; do | |
| if [ -d "$dir" ]; then | |
| provider=$(basename "$dir" | sed 's/eval-results-//') | |
| file="$dir/eval-results-${provider}.json" | |
| if [ -f "$file" ]; then | |
| passed=$(jq '[.results[] | select(.passed == true)] | length' "$file") | |
| failed=$(jq '[.results[] | select(.passed == false)] | length' "$file") | |
| total=$(jq '.results | length' "$file") | |
| echo "| $provider | $passed | $failed | $total |" >> $GITHUB_STEP_SUMMARY | |
| fi | |
| fi | |
| done | |
| echo "" >> $GITHUB_STEP_SUMMARY | |
| echo "Run at: $(date -u)" >> $GITHUB_STEP_SUMMARY |