feat: Copilot SDK migration + E2E test consolidation#25
Merged
Conversation
- Replace Vercel AI SDK + Azure AI Foundry with GitHub Copilot SDK - Add Node.js proxy server (src/server.js + src/copilotProxy.js) that spawns @github/copilot CLI and bridges it via WebSocket to the task pane - Add browser WebSocket client (src/lib/websocket-client.ts + websocket-transport.ts) implementing Copilot LSP over JSON-RPC - Rewrite useOfficeChat.ts: WebSocket session lifecycle + useExternalStoreRuntime - Convert tool factory from Zod ToolConfig to Copilot SDK Tool[] (JSON Schema) - Remove SetupWizard, SettingsDialog, aiClientFactory, chatService, modelDiscoveryService, providerConfig -- auth is now GitHub account-based - Simplify settingsStore: remove endpoint/model CRUD, add activeModel with COPILOT_MODELS list; bump persist version to 2 (clean break) - Delete all live-API integration tests; keep jsdom unit + component tests - All 259 tests pass, tsc and webpack build clean - Update README and CONTRIBUTING to reflect new architecture Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Three tests covering the full WebSocket → proxy → Copilot API path: 1. Connect to proxy server 2. Create a session and get a real LLM response (checks assistant.message_delta) 3. Execute a tool call and verify the handler fires Tests skip automatically when npm run server is not running. Uses SystemMessageConfig typed correctly (mode: 'append'). Also updated vitest.integration.config.ts: - Switch to jsdom environment (WebSocket global available) - Include .test.tsx files - Drop stale Foundry dotenv load Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
…al server - Switch to node environment (// @vitest-environment node) so real TCP connections work (jsdom's WebSocket is a stub) - Patch global WebSocket with ws package (rejectUnauthorized: false) so Node trusts the office-addin-dev-certs localhost cert - Handle both assistant.message and assistant.message_delta event types (Copilot emits complete message, not streaming deltas, in this flow) - Fix SERVER_URL to wss:// (server is HTTPS/WSS) All 3 tests now pass against live GitHub Copilot: ✓ connects to the proxy server ✓ creates a session and gets a response to a simple prompt ✓ executes a tool call and returns the result to the model Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
- useOfficeChat: use type guard predicate instead of non-null assertion - websocket-client: use undefined-check instead of non-null assertion - server.js: add eslint-disable comments for intentional console logs - ci.yml: remove Azure/Foundry secrets, integration tests always run (Copilot WebSocket tests skip gracefully when no server available) Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Integration tests were missing two critical settings vs the unit config: - setupFiles: ['tests/setup.ts'] — needed for OfficeRuntime mock and jsdom polyfills (ResizeObserver, matchMedia, IntersectionObserver) - globals: true — needed for @testing-library/react auto-cleanup; without it, afterEach is not global and cleanup never runs, causing stale Radix UI portal elements to accumulate across tests Result: 47/47 integration tests passing (up from 12/47) Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
- .github/copilot-instructions.md: full rewrite for Copilot SDK architecture (WebSocket proxy diagram, updated unit/integration test tables with 18/12 files, new key files list, OfficeRuntime test guidance) - README.md: replace Chat Architecture, Key Hooks, Authentication, Tech Stack sections to reflect Copilot SDK + proxy server - docs/TESTING_STRATEGY.md: full rewrite with current test file tables, correct counts, removed Azure/Foundry references Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
- tests-e2e/src/test-taskpane.ts: replace Azure/Foundry AI round-trip with GitHub Copilot WebSocket client (createWebSocketClient + tool.execution_start events); skips gracefully when proxy unavailable - tests-e2e/webpack.config.js: replace FOUNDRY_* DefinePlugin vars with COPILOT_SERVER_URL; remove Entra ID token acquisition - tests-ui/fixtures.ts: update makeSettingsJSON to match current UserSettings shape (activeModel, no endpoints); fix stale comments - tests-ui/chat/chat-ui.spec.ts: update model name gpt-4.1 → Claude Sonnet 4.5 - tests-ui/chat/chat-copilot.spec.ts: new Playwright E2E test that sends a message and verifies AI response; auto-skips when server unavailable Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
- Add 11 alternate tool name aliases (find_values, insert_range, delete_range, merge_cells, unmerge_cells, group_rows_columns, ungroup_rows_columns, auto_fill_range, get_range_precedents, get_range_dependents, get_tables_for_range) - Fix copy_range: rename sourceAddress->address, add copiedTo result field - Fix filter_table: rename values->filterValues arg - Fix resize_table: rename newAddress->address arg - Add set_table_style, set_table_header_totals_visibility mappings - Fix set_chart_title: include title in result transform - Add set_chart_data_source mapping - Fix edit_comment: rename newText->text arg - Fix CF add_* operations: add applied result transform for all - Fix add_data_bar: rename barColor->fillColor - Fix add_cell_value_format: rename fillColor->backgroundColor - Fix add_top_bottom_format: rename rank->topBottomRank, topOrBottom->topBottomType, fillColor->backgroundColor; add defaults for rank/type - Fix add_color_scale:defaults: provide default min/max colors - Fix add_text_contains_format/add_contains_text_format: rename text->containsText, fillColor->backgroundColor - Fix add_custom_format: rename formula->formula1, fillColor->backgroundColor - Fix sort_pivot_field_values: include valuesHierarchyName in result - Fix get_pivot_table_location: use rangeAddress field name - Fix refresh_all_pivot_tables: return refreshed:true field - Fix set_number_validation: support decimal type (maps to Excel decimal rule) - Add decimal type support to dataValidation.config.ts All 230 E2E tests now passing (4 pending AI round-trip tests) Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
- Update Playwright fixtures and UI test specs to match current app state - Simplify wizard-flow.spec.ts (remove SetupWizard/SettingsDialog tests no longer applicable) - Add useCustomSortLists param to pivotTable.config.ts configure action - All 15 UI tests pass, 230 E2E tests pass, 265 unit/integration tests pass Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
# Conflicts: # src/tools/configs/dataValidation.config.ts # src/tools/configs/pivotTable.config.ts # src/tools/index.ts # tests/unit/toolSchemas.test.ts
sbroenne
pushed a commit
that referenced
this pull request
Feb 20, 2026
This reverts commit 5596abe.
sbroenne
pushed a commit
that referenced
this pull request
Feb 20, 2026
This reverts commit aaf2100.
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
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
All tests passing: 265 unit/integration, 15 UI, 230/234 E2E (4 pending need live Copilot). Tool consolidation 134->10 configs. LEGACY_TOOL_MAP for E2E compat.
Co-authored-by: Copilot 223556219+Copilot@users.noreply.github.com