🤖 Add workflow_dispatch for Terminal-Bench (#290)

_Generated with `cmux`_

Adds a manually-triggerable GitHub Actions workflow for running `make
benchmark-terminal`.

**Features:**
- Workflow can be triggered from GitHub Actions UI with custom
parameters
- 3 hour timeout to accommodate long-running benchmarks
- Installs `uv` for `uvx terminal-bench` execution
- Configurable inputs:
  - Dataset (default: `terminal-bench-core==0.1.1`)
  - Concurrency (default: `4`)
  - Livestream (default: `true` for progress visibility)
  - Extra args for custom options
- Uploads benchmark results as artifacts (even on failure)
- Uses ANTHROPIC_API_KEY and OPENAI_API_KEY from secrets

Author: ammar-agent

.github/workflows/ci.yml

@@ -46,6 +46,12 @@ jobs:
           extra_nix_config: |
             experimental-features = nix-command flakes
 
+      - name: Install uv
+        run: curl -LsSf https://astral.sh/uv/install.sh | sh
+
+      - name: Add uv to PATH
+        run: echo "$HOME/.local/bin" >> $GITHUB_PATH
+
       - name: Run static checks
         run: make -j3 static-check
 

.github/workflows/terminal-bench.yml

@@ -0,0 +1,67 @@
+name: Terminal-Bench
+
+on:
+  workflow_dispatch:
+    inputs:
+      dataset:
+        description: 'Terminal-Bench dataset to use'
+        required: false
+        default: 'terminal-bench-core==0.1.1'
+        type: string
+      concurrency:
+        description: 'Number of concurrent tasks (--n-concurrent)'
+        required: false
+        default: '4'
+        type: string
+      livestream:
+        description: 'Enable livestream mode'
+        required: false
+        default: true
+        type: boolean
+      extra_args:
+        description: 'Additional arguments to pass to terminal-bench'
+        required: false
+        type: string
+
+jobs:
+  benchmark:
+    name: Run Terminal-Bench
+    runs-on: ${{ github.repository_owner == 'coder' && 'depot-ubuntu-22.04-16' || 'ubuntu-latest' }}
+    timeout-minutes: 180 # 3 hours - terminal-bench can take a long time
+    steps:
+      - name: Checkout code
+        uses: actions/checkout@v4
+        with:
+          fetch-depth: 0 # Required for git describe to find tags
+
+      - uses: ./.github/actions/setup-cmux
+
+      - name: Install uv
+        run: curl -LsSf https://astral.sh/uv/install.sh | sh
+
+      - name: Add uv to PATH
+        run: echo "$HOME/.local/bin" >> $GITHUB_PATH
+
+      - name: Generate version file
+        run: ./scripts/generate-version.sh
+
+      - name: Run Terminal-Bench
+        run: make benchmark-terminal
+        env:
+          TB_DATASET: ${{ inputs.dataset }}
+          TB_CONCURRENCY: ${{ inputs.concurrency }}
+          TB_LIVESTREAM: ${{ inputs.livestream && '1' || '' }}
+          TB_ARGS: ${{ inputs.extra_args }}
+          ANTHROPIC_API_KEY: ${{ secrets.ANTHROPIC_API_KEY }}
+          OPENAI_API_KEY: ${{ secrets.OPENAI_API_KEY }}
+
+      - name: Upload benchmark results
+        if: always()
+        uses: actions/upload-artifact@v4
+        with:
+          name: terminal-bench-results
+          path: |
+            terminal-bench-results/
+            *.json
+          if-no-files-found: warn
+

benchmarks/__init__.py

@@ -3,23 +3,25 @@
 # This file contains all code formatting logic.
 # Included by the main Makefile.
 
-.PHONY: fmt fmt-check fmt-prettier fmt-prettier-check fmt-shell fmt-shell-check fmt-nix fmt-nix-check
+.PHONY: fmt fmt-check fmt-prettier fmt-prettier-check fmt-shell fmt-shell-check fmt-nix fmt-nix-check fmt-python fmt-python-check
 
 # Centralized patterns - single source of truth
 PRETTIER_PATTERNS := 'src/**/*.{ts,tsx,json}' 'tests/**/*.ts' 'docs/**/*.md' 'package.json' 'tsconfig*.json' 'README.md'
 SHELL_SCRIPTS := scripts
+PYTHON_DIRS := benchmarks
 
 # Always use bun x prettier for reproducibility (uses package.json version)
 PRETTIER := bun x prettier
 
 # Tool availability checks
 SHFMT := $(shell command -v shfmt 2>/dev/null)
 NIX := $(shell command -v nix 2>/dev/null)
+UVX := $(shell command -v uvx 2>/dev/null)
 
-fmt: fmt-prettier fmt-shell fmt-nix
+fmt: fmt-prettier fmt-shell fmt-python fmt-nix
 	@echo "==> All formatting complete!"
 
-fmt-check: fmt-prettier-check fmt-shell-check fmt-nix-check
+fmt-check: fmt-prettier-check fmt-shell-check fmt-python-check fmt-nix-check
 	@echo "==> All formatting checks passed!"
 
 fmt-prettier:
@@ -48,6 +50,21 @@ else
 	@shfmt -i 2 -ci -bn -d $(SHELL_SCRIPTS)
 endif
 
+# Helper target to check for uvx
+.check-uvx:
+ifeq ($(UVX),)
+	@echo "Error: uvx not found. Install with: curl -LsSf https://astral.sh/uv/install.sh | sh"
+	@exit 1
+endif
+
+fmt-python: .check-uvx
+	@echo "Formatting Python files..."
+	@uvx ruff format $(PYTHON_DIRS)
+
+fmt-python-check: .check-uvx
+	@echo "Checking Python formatting..."
+	@uvx ruff format --check $(PYTHON_DIRS)
+
 fmt-nix:
 ifeq ($(NIX),)
 	@echo "Nix not found; skipping Nix formatting"