Visit files and run shell commands from chat #684

	name: Integration Tests

	on:
	push:
	branches: [master, main]
	pull_request:

	jobs:
	fake:
	name: Fake contract
	runs-on: ubuntu-latest
	timeout-minutes: 10
	steps:
	- uses: actions/checkout@v4
	- uses: ./.github/actions/setup-emacs
	with:
	version: '30.1'
	- name: Run fake backend contract
	run: make test-integration-fake

	real:
	name: Real contract
	runs-on: ubuntu-latest
	timeout-minutes: 20
	services:
	ollama:
	image: ollama/ollama:0.11.11
	ports:
	- 11434:11434
	steps:
	- uses: actions/checkout@v4
	- uses: ./.github/actions/setup-emacs
	with:
	version: '30.1'
	- uses: actions/setup-node@v4
	with:
	node-version: '24'
	- name: Setup Ollama model
	run: \|
	# Wait for Ollama to be ready
	for i in {1..30}; do curl -s http://localhost:11434/api/tags && break; sleep 1; done

	# Pull the model
	echo "Pulling qwen3:1.7b..."
	curl -s http://localhost:11434/api/pull -d '{"name":"qwen3:1.7b"}' \| tail -1

	# Verify
	echo "Verifying model..."
	curl -s http://localhost:11434/api/tags \| grep -q 'qwen3:1.7b' \|\| {
	echo "::error::Model not found after pull"
	curl -s http://localhost:11434/api/tags
	exit 1
	}
	echo "Model qwen3:1.7b is ready"

	# Warm up: first inference loads weights into memory (10-30s cold start)
	echo "Warming up model..."
	curl -s http://localhost:11434/api/generate \
	-d '{"model":"qwen3:1.7b","prompt":"hi","stream":false,"think":false,"options":{"num_predict":1}}' > /dev/null
	echo "Warmup complete"
	- name: Run real backend contract
	run: make test-integration-ci-real
	env:
	PI_CODING_AGENT_DIR: ${{ runner.temp }}/pi-test

Provide feedback