All notable changes to On-Call Copilot are documented here.
- Starlette compatibility pin (
requirements.txt): Addedstarlette>=0.36.0,<1.0.0to prevent startup failure when pip resolvesstarlette==1.0.0rc1.azure-ai-agentserver-core==1.0.0b14uses theon_event()decorator API that was removed in Starlette 1.0. Without this pin the application exits at startup withAttributeError: 'Starlette' object has no attribute 'on_event'.
Hosting_Agent.md: Major revision — 360 lines added, 156 removed. Updated hosting and deployment guidance for the Foundry hosted-agent workflow.
README.md: Updated branding references from Azure AI Foundry to Microsoft Foundry.
AGENTS.md: Full multi-agent architecture reference added (255 lines) — documents all four specialist agents (Triage, Summary, Comms, PIR), their output schemas, orchestration viaConcurrentBuilder, and the output merge contract.Hosting_Agent.md: New guide (604 lines) covering end-to-end deployment of the hosted agent to Microsoft Foundry including Docker, ACR, andagent-dev-cliworkflows.scripts/test_all_demos.py: New script (114 lines) that runs all 8 incident scenarios (3 demos + 5 scenarios) against the UI server.scripts/test_one.py: New helper script (19 lines) for running a single scenario..env.example: Added 47-line example environment file documenting all required environment variables..vscode/mcp.json: Added VS Code MCP server configuration.
CONTRIBUTING.md: Extended contributing guide with additional workflow detail.scripts/invoke.py: Updated invocation script with revised argument handling.scripts/scenarios/README.md: Updated scenario documentation.scripts/scenarios/scenario_4_minimal_alert.json: Updated minimal alert scenario payload.ui/server.py: Extended UI server with improved request handling and endpoint configuration.docs/blog_post.md: Revised and expanded blog post content.app/main.py: Minor additions to application startup.main.py: Minor additions to entry-point startup.requirements.txt: Dependency updates.README.md: Comprehensive rewrite — restructured setup, configuration, testing, and architecture sections. Consolidated and compressed existing content.docs/architecture.excalidraw: Architecture diagram updated.docs/screenshots/: Screenshot assets reorganised — old per-panel screenshots underui/subfolder removed; consolidated set published at top-levelscreenshots/.
ui/index.html: Full web UI (1 479 lines) — four-panel results view with Triage, Summary, Comms, and PIR panels; Quick Load presets for 3 demos and 5 scenarios; JSON editor with validation; live status bar.ui/server.py: Static-file and proxy server (258 lines) for the web UI usingInteractiveBrowserCredential.scripts/make_demo_video.py: Automated demo-video generation script (397 lines) using Playwright.docs/blog_post.md: Technical blog post describing the architecture and design decisions.docs/screenshots/: Full screenshot set for README and documentation (17 UI screenshots + model router and model selection screenshots).docs/demo_ui.mp4: Recorded demo video.
README.md: Major expansion — added UI walkthrough, Foundry Playground screenshots, architecture diagram updates, and additional setup instructions.docs/architecture.excalidraw: Architecture diagram substantially revised.
First public commit establishing the full project:
- Four specialist agents (
app/agents/):triage.py,summary.py,comms.py,pir.py— each with its own system-prompt instructions and designated JSON output keys. - Orchestrator (
main.py,app/main.py):ConcurrentBuilderwiring all four agents viaasyncio.gather(); JSON fragment merge with telemetry injection. - Schema validation (
app/schemas.py): Pydantic models for all agent output keys. - Mock router (
app/mock_router.py): LocalMOCK_MODEresponses for schema-only validation without Azure. - Prompting utilities (
app/prompting.py), telemetry (app/telemetry.py). - Agent definition (
agent.yaml): Declarative hosted-agent registration for Foundry Responses API on port 8088. - Infrastructure (
infra/main.bicep,azure.yaml): Minimal Bicep stub andazdconfiguration. - Dockerfile: Multi-stage container build.
- Scripts:
invoke.py,run_scenarios.py,validate.py,verify_agent.py,deploy_sdk.py,get_logs.py, local test helpers (test_local.http,test_local.ps1,test_local.sh). - Scenarios: 5 JSON incident scenarios + 3 demos; 5 golden output files for schema validation.
- Documentation:
README.md,CONTRIBUTING.md,SECURITY.md,docs/CONFIGURATION.md,scripts/SCENARIOS.md.