Commit Graph
100 Commits
Author SHA1 Message Date
David KayaandCopilot e38a663834 fix: fall back to SSE transport when Streamable HTTP probe fails
Many MCP servers only support legacy SSE despite being configured as
generic HTTP endpoints. Follow the MCP SDK's recommended fallback
pattern: try Streamable HTTP first, then retry with SSE on failure.
This matches VS Code's behavior for these servers.

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
2026-03-28 19:39:23 +01:00
David KayaandCopilot f0114058ba fix: show MCP tools in all provider groups when shared
When multiple MCP servers expose the same tool names (e.g., several
kusto-mcp instances), each server's group now shows its tools instead
of only the first server getting the tools and others showing 0/0.

Deduplicate the effectiveAutoApprovedCount to avoid double-counting
shared tools in the pill header.

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
2026-03-28 19:35:56 +01:00
David KayaandCopilot 6505493735 fix: resolve nested button DOM nesting in approval pill
Change the approval group header from <button> to a <div> with
role='button' and keyboard handling to avoid nesting the GroupToggle
<button> inside another <button>, which is invalid HTML.

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
2026-03-28 19:33:41 +01:00
David KayaandCopilot b4b0bf54d2 fix: improve MCP tool probing reliability
- Increase default probe timeout from 10s to 30s to handle slow
  package managers (uvx, npx) that install on first run
- Add console logging for probe success/failure to aid debugging
- Probe manually configured MCP servers (not just discovered ones)
  so all servers with wildcard tools get their tools discovered

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
2026-03-28 19:30:42 +01:00
David KayaandCopilot 8312a47bf1 feat: discover MCP tools via protocol probing for approval pill
When MCP server configs declare wildcard tools (empty tools array),
the Copilot SDK has no API to list individual tool names. Add direct
MCP protocol probing using @modelcontextprotocol/sdk to discover
available tools from each accepted server.

- Add McpToolProber service supporting stdio/SSE/HTTP transports
- Probe accepted MCP servers on project load, acceptance, and rescan
- Store probed tools on DiscoveredMcpServer and McpServerDefinition
- Use probed tools in listApprovalToolDefinitions when declared tools
  are empty, so the approval pill shows individual tool toggles
- Remove unused isMcpServerApprovalKey helper
- Fix effectiveAutoApprovedCount Math.max workaround in ChatPane
- Add comprehensive tests for probed tool behavior
- Update ARCHITECTURE.md tooling integration section

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
2026-03-28 19:23:57 +01:00
David KayaandCopilot 3937904548 feat: add server-level MCP auto-approval for wildcard tool servers
MCP servers configured with empty tools arrays (wildcard) now appear in the
auto-approval pill with a server-level toggle. When toggled, a server-level
approval key (mcp_server:<name>) is added to autoApprovedToolNames. The
sidecar matches this key against PermissionRequestMcp.ServerName to auto-
approve all tools from that server without needing individual tool names.

- Add buildMcpServerApprovalKey/isMcpServerApprovalKey helpers
- Create approval groups for all MCP servers including empty-tools ones
- Add serverApprovalKey to ApprovalToolGroup for server-level toggles
- Update sidecar RequiresToolCallApproval to check server-level keys
- Include server-level keys in listApprovalToolNames for pruning safety
- Add tests for both shared domain and sidecar approval matching

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
2026-03-28 18:34:19 +01:00
David KayaandCopilot 53a08e0ed4 feat: redesign auto-approval pill with server-level grouping and batch toggle
Group MCP tools by server and LSP tools by profile in the approval popover
instead of showing a flat list. Each server/profile group gets a collapsible
header with a batch toggle to approve/unapprove all tools at once. Add a
search filter (visible when >10 tools) and partial-state indicators for
groups with mixed approval. Add groupApprovalToolsByProvider shared helper
with tests.

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
2026-03-28 18:18:17 +01:00
David KayaandCopilot 5ad85db0f5 refactor: redesign project settings with sidebar navigation
Replace flat scrolling layout with sidebar + content panel matching the global
SettingsPanel pattern. Sections: Overview, Instructions (with expandable
previews), Custom Agents (with enable/disable), Prompt Files, MCP Servers,
Danger Zone. Add count badges, pending indicators, project name in header,
section-level rescan buttons, and contextual empty states.

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
2026-03-28 17:59:31 +01:00
David KayaandCopilot dd203ddde5 feat: add frontend customization UI and prompt picker
Add Copilot Customization section to ProjectSettingsPanel with instruction file
previews, agent profile enable/disable toggles, and prompt file listing. Add
InlinePromptPill to chat input for selecting and sending prompt files with
variable substitution. Wire rescan and agent profile IPC in App.tsx. Update
website with Copilot Customization feature card.

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
2026-03-28 17:40:21 +01:00
David KayaandCopilot 75b9ff667a feat: support project copilot customization
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
2026-03-28 17:29:37 +01:00
David KayaandCopilot f53907755a feat: add dedicated project settings panel accessible from sidebar
Move project-specific settings (discovered MCP servers) out of the
global Settings panel and into a dedicated ProjectSettingsPanel overlay.
Users can now access any project's settings via a gear icon on the
sidebar project header, or by clicking the pending-discovery badge.

- Add ProjectSettingsPanel component with project info, discovered MCP
  server management (accept/dismiss/rescan), and project removal
- Add gear icon and clickable discovery badge to Sidebar ProjectGroup
- Remove project-specific discovery props from SettingsPanel (now
  shows only user-level discovered servers)
- Wire ProjectSettingsPanel overlay in App.tsx with auto-close on
  project removal

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
2026-03-28 16:51:57 +01:00
David KayaandCopilot 3c57cb6ded fix: clean up interrupted session state on restart
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
2026-03-28 16:22:12 +01:00
David KayaandCopilot b946359c69 fix: stabilize macOS CI validation
Remove the packaging-only create-dmg dependency in favor of native hdiutil, and make the hook runner working-directory test assert behavior without depending on macOS temp-path canonicalization.

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
2026-03-28 15:43:13 +01:00
David Kaya e4eb221308 fix: new screenshot for website 2026-03-28 15:30:41 +01:00
David KayaandCopilot 036fb4d4fa fix: stabilize cross-platform CI checks
Use the repository's default Electron import pattern in the MCP OAuth service, make the attachment-path test platform-neutral, and ensure the hook runner cwd/env test drains stdin before asserting shell output.

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
2026-03-28 15:27:48 +01:00
David KayaandCopilot 6321f9192d docs(website): add missing feature cards and expand existing copy
Add 5 new feature cards: Visual Pattern Editor, Extensible Tooling, Tooling
Discovery, Plan Review & Agent Questions, and Run Timeline. Expand Real
Project Context (git awareness), Persistent Sessions (search/duplicate),
and Get Started step 1 (Copilot diagnostics). Enrich tech banner with
React Flow and Open Source. Feature grid goes from 8 to 13 cards.

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
2026-03-28 14:52:48 +01:00
David KayaandCopilot 147b437e36 fix: isolate scratchpad session working directories
Give each scratchpad session its own working directory, migrate existing scratchpad sessions on workspace load, and route sidecar turns through the session-specific cwd. Add regression coverage for create, duplicate, delete, and migration flows, and document the new persistence model.

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
2026-03-28 14:31:39 +01:00
David KayaandCopilot fa5774cbc0 docs: strengthen backend/frontend plan-splitting requirement in AGENTS.md
Make the two-phase planning rule a hard, no-exceptions requirement.
Clarify sequencing, separate-session expectation, and handover
artifact contract coverage. Add when-in-doubt-split directive.

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
2026-03-28 14:08:30 +01:00
David KayaandCopilot ae56d55c85 fix: clean up hook event labels in Activity panel
Replace raw SDK identifiers with human-friendly labels (e.g.
sessionStart → Session start) and drop the meaningless UUID detail.
Labels now read like 'Session start hook started' instead of
'Hook sessionStart started'.

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
2026-03-28 14:06:21 +01:00
David KayaandCopilot 1fbfdbbac4 fix: suppress noisy hook lifecycle events
Stop forwarding hook lifecycle events when a project has no configured
.github/hooks/*.json commands so the UI only receives meaningful hook
activity. Also expose the configured-hook state on the bundle, cover the
suppression wiring with tests, and document the updated event semantics.

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
2026-03-28 14:05:10 +01:00
David KayaandCopilot db9ffe8399 fix: disambiguate hook lifecycle events in Activity panel
Append phase text (started/completed) to hook event labels so
start and end phases are visually distinct instead of appearing
as duplicates.

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
2026-03-28 13:25:35 +01:00
David KayaandCopilot dff97efd58 feat: support project hook commands
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
2026-03-28 13:23:01 +01:00
David KayaandCopilot f459cc7291 fix: stop forcing copilot session ids
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
2026-03-28 12:52:52 +01:00
David KayaandCopilot e13e3b818a fix: restore final assistant responses
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
2026-03-28 12:44:14 +01:00
David KayaandCopilot f1fa52f9c3 feat: full Copilot SDK feature parity — custom agents, hooks, image input, skills, steering, session persistence
Backend (sidecar):
- Extended ProtocolModels with DTOs for custom agents, hooks, skills,
  infinite sessions, session lifecycle, and 9 new event types
- Added CopilotManagedSessionIds for stable SDK session ID mapping
- Added CopilotSessionManager/ICopilotSessionManager for session lifecycle
- Added CopilotSessionHooks for hook registration
- Added CopilotMessageOptionsMetadata for mid-turn steering
- Extended CopilotAgentBundle to wire custom agents, hooks, skills,
  infinite sessions, and stable session IDs
- Extended CopilotTurnExecutionState to project 13 new SDK event types
- Widened ITurnWorkflowRunner callback to accept SidecarEventDto
- Added list/delete/disconnect session commands to SidecarProtocolHost
- Added AryxCopilotAgentMessageOptionsTests (14 new tests, 142 total)

Frontend (renderer + main + shared):
- Added ChatMessageAttachment type and helpers (attachment.ts)
- Extended sidecar contracts with MessageMode, 3 new command types,
  9 new event types, and agent/session config DTOs
- Extended SessionEventRecord with 6 new event kinds and ~20 fields
- Added PatternAgentCopilotConfig to pattern domain
- Added attachments support to ChatMessageRecord
- Updated sidecar client with session lifecycle methods and
  turn-scoped event routing via onTurnScopedEvent callback
- Updated main process: handleTurnScopedEvent(), deleteSession(),
  steering bypass for mid-turn messages, attachment passthrough
- Added deleteSession IPC handler and preload binding
- Added TurnEventLog state tracker with format/apply/prune helpers
- ChatPane: always-enabled composer, steering indicator, attachment
  picker with preview, image thumbnails in message history,
  context-usage bar, amber steer mode for send button
- ActivityPanel: turn events section with sub-agent, hook, skill,
  and compaction event rendering
- Sidebar: delete session action in context menu
- App.tsx: wired sessionUsage, turnEventLogs, and deleteSession

Documentation:
- AGENTS.md: added glob safety rule for node_modules
- README.md: added steering, image input, and richer observability
- ARCHITECTURE.md: added turn-scoped events, steering, and attachments
- Website: added steering and image input feature cards, updated
  live visibility and session cards

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
2026-03-28 12:28:20 +01:00
David Kaya 0c2973c599 fix: add plan to .gitignore 2026-03-28 11:54:32 +01:00
David KayaandCopilot c1dab96bfd fix: wire PatternEditor to mode-aware graph mutations
Replace the old disconnected-node addAgentNodeToGraph with the
mode-aware addAgentToGraph so new agents are automatically wired
into the graph topology. Replace the graph-nuking removeAgent
path with removeAgentFromGraph for surgical node removal. Split
emitChange into graph-preserving and mode-change paths so editing
agent properties no longer discards custom positions and edges.

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
2026-03-27 21:55:54 +01:00
David KayaandCopilot bb7e5d4108 fix: preserve editable pattern graphs
Add mode-aware graph mutation helpers for pattern topology so backend
code can add and remove agents without forcing a full graph rebuild.
Also preserve an existing pattern graph in savePattern instead of
re-syncing it from defaults, and cover the new behavior with shared
and main-process tests.

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
2026-03-27 21:50:01 +01:00
David KayaandCopilot b5ab92e444 fix: auto-resolve queued approvals of the same category
When multiple tool calls of the same permission category (e.g. three
concurrent view/read calls) arrive simultaneously, approving one now
cascades to all queued approvals sharing the same category key. This
eliminates repeated approval prompts for parallel tool calls.

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
2026-03-27 21:43:21 +01:00
David KayaandCopilot 689b335220 feat: category-based approval for runtime tools
Replace 20+ individual runtime tool toggles with 5 permission categories
(read, write, shell, web_fetch, store_memory) matching how Copilot CLI,
Cline, and other coding agents handle approvals.

- Add resolveApprovalToolKey() to map permission kinds to category IDs
- Simplify fallbackRuntimeApprovalTools to 5 categories
- Thread alwaysApprove through PendingApprovalHandle → sidecar resolve
- Update ApprovalBanner to use category labels for Always Approve button
- Update ResolveApprovalCommand contract with alwaysApprove field
- Update tests for category-based model

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
2026-03-27 21:36:27 +01:00
David KayaandCopilot f04e3b9dcc feat: cache same-turn approval selections
Add an AlwaysApprove flag to resolve-approval commands and cache
approved tool keys for the current request so repeated runtime
permissions are auto-approved immediately within the same turn.
The cache is cleared when the turn finishes to avoid leaking across
future requests.

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
2026-03-27 21:24:55 +01:00
David KayaandCopilot 3ec69d990b feat: enable 'always approve for session' for runtime tools
Runtime tool permission requests (read, write, shell, store_memory) now
resolve stable approval names via fallback in the sidecar approval
coordinator, enabling the 'Always approve' button and session-level
auto-approval for built-in tools.

Backend:
- Add fallback tool names for PermissionRequestRead (read),
  PermissionRequestWrite (write), PermissionRequestShell (shell),
  and PermissionRequestMemory (store_memory)
- Add autoApprovedToolName parameter to RequiresToolCallApproval
  for permission-kind-based session approval matching
- Track tool.execution_start events in ToolNamesByCallId for
  supplementary exact-name resolution

Frontend:
- Use approval.toolName ?? approval.permissionKind as fallback key
  in resolveSessionApproval and ApprovalBanner
- Add shell, read, write permission-kind entries to builtin approval
  tool definitions so pruning preserves session overrides

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
2026-03-27 21:20:14 +01:00
David KayaandCopilot 154787c336 fix: only project handoff tool calls as FunctionCallContent
AryxCopilotAgent was converting ALL Copilot tool requests (ask_user,
MCP tools, etc.) into FunctionCallContent. The Agent Framework's
AIAgentHostExecutor tracks every FunctionCallContent as an outstanding
request.  Since SDK-handled tools like ask_user never produce a
matching FunctionResultContent, HasOutstandingRequests stayed true and
TurnToken was never emitted — stalling group-chat advancement after
an ask_user interaction.

Now only handoff_to_* tool requests are projected as FunctionCallContent.
All other tool calls remain invisible to the Agent Framework executor,
matching the upstream GitHubCopilotAgent behaviour.

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
2026-03-27 21:12:42 +01:00
David KayaandCopilot 9d65e3d209 fix: restore real Copilot handoffs
Replace the broken forced-tool workaround with a repo-local Copilot
agent adapter that merges runtime handoff instructions and tool
declarations into Copilot sessions and projects Copilot tool requests
back into Agent Framework function-call updates.

Also add regression coverage for the adapter and document the runtime
integration detail in the architecture guide.

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
2026-03-27 21:07:58 +01:00
David KayaandCopilot 7247a68f24 fix: enforce real entry handoffs
Constrain the initial handoff triage agent so it cannot use regular
session tooling and must use a handoff tool on its first invocation.
This prevents the model from narrating delegation in plain text while
never actually transferring control.

Also add regression tests for the new entry-agent constraints.

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
2026-03-27 20:58:31 +01:00
David Kaya 85a9327f65 fix: surface handoff specialist activity\n\nQueue per-agent thinking updates from synchronous Copilot SDK session\nevents, flush them through the workflow runner, and promote handoff\ntargets to thinking immediately after handoff activity is emitted.\nThis fixes the UI case where downstream agents stayed on 'No status\nyet' during handoff patterns.\n\nAlso add targeted handoff diagnostics and regression tests.\n\nCo-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> 2026-03-27 20:49:52 +01:00
David KayaandCopilot 2a952dfebe fix: workspace reset now properly clears UI state
resetLocalWorkspace() deleted workspace.json and the scratchpad
directory but never broadcast the fresh state to the renderer.
The renderer also discarded the IPC return value. This left stale
projects, sessions, and activities visible in the UI after reset.

- Emit workspace-updated after reloading the workspace in the backend
- Use the returned fresh workspace to immediately update renderer state
- Clear sessionActivities and close the settings panel on reset

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
2026-03-27 20:48:37 +01:00
David KayaandCopilot cf41279ff5 fix: clamp auto-approval pill count to known tools
The auto-approval pill displayed effectiveAutoApproved.size as the
numerator, which could include tool names not present in the approval
tool list (e.g. names added via 'Always approve' for tools that aren't
in runtimeTools yet). This caused impossible counts like '17/14'.

Compute the count by intersecting effectiveAutoApproved with the actual
approvalTools list, so the numerator never exceeds the denominator.

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
2026-03-27 20:40:30 +01:00
David KayaandCopilot c02589f4c0 fix: only probe OAuth for newly enabled MCP servers
Previously, every tooling update re-probed all enabled HTTP MCP servers
for OAuth requirements. Now it compares the new selection against the
previous one and only probes servers that were just enabled.

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
2026-03-27 20:31:29 +01:00
David KayaandCopilot 73595039fc fix: support provider-specific MCP OAuth settings
Use provider-aware OAuth handling instead of a single hardcoded client ID.
GitHub now follows VS Code's GitHub auth setup with the GitHub client ID,
GitHub authorize/token endpoints, prompt=select_account, vscode.dev/redirect,
and JSON-compatible token exchange. Entra keeps the Copilot client ID via a
known provider match, while unknown providers still fall back to metadata
and dynamic registration.

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
2026-03-27 20:23:26 +01:00
David KayaandCopilot 81bddcbd63 fix: align MCP OAuth flow with Copilot client ID requirements
- Use vscode.dev/redirect as redirect URI when using the Copilot client
  ID, with the local callback URL encoded in the state parameter
- Use PRM resource scopes (e.g. api://icmmcpapi-prod/mcp.tools) instead
  of generic OIDC scopes, always include offline_access
- Listen on root path instead of /callback to match vscode.dev/redirect
  behavior
- Extract PRM scopes from protected resource metadata discovery

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
2026-03-27 20:15:25 +01:00
David KayaandCopilot 53f1167681 feat: use GitHub Copilot client ID as default for MCP OAuth
Use the well-known Copilot client ID (aebc6443-996d-45c2-90f0-388ff96faa56)
when no static client config is provided and the auth server doesn't
support dynamic client registration. This matches how VS Code authenticates
with Entra ID-backed MCP servers.

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
2026-03-27 20:08:40 +01:00
David KayaandCopilot 4f7b479996 fix: fall back to openid-configuration for auth server metadata discovery
Microsoft Entra ID and other OIDC providers expose metadata at
.well-known/openid-configuration rather than RFC 8414's
.well-known/oauth-authorization-server. Now tries both suffixes.

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
2026-03-27 20:06:02 +01:00
David KayaandCopilot 0fd7a04a51 fix: try both RFC 9728 and appended well-known URL paths for MCP OAuth discovery
Some MCP servers place .well-known metadata at the appended path
(e.g. /v1/.well-known/oauth-protected-resource) instead of the
RFC 9728 compliant inserted path. Now tries the RFC-compliant URL
first and falls back to the appended form.

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
2026-03-27 20:00:50 +01:00
David KayaandCopilot 6c6b49fde4 fix: probe PRM directly instead of GET-ing the MCP server URL
MCP servers use POST/SSE, not GET. A plain GET to the server URL can
return 405, 404, or other codes unrelated to authentication. Instead,
go directly to the .well-known/oauth-protected-resource endpoint and
check whether it returns valid metadata with authorization_servers.

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
2026-03-27 19:48:49 +01:00
David KayaandCopilot d73eaae30b fix: correct well-known URL construction per RFC 9728
The .well-known/{suffix} segment was being appended to the full URL
path instead of inserted after the origin. For example:

  Wrong:   https://api.example.com/mcp/.well-known/oauth-protected-resource
  Correct: https://api.example.com/.well-known/oauth-protected-resource/mcp/

- Extract buildWellKnownUrl() helper in mcpTokenStore (Electron-free)
- Fix all 3 call sites: requiresOAuth probe, PRM discovery, auth
  server metadata
- Add 6 unit tests for URL construction with various path shapes

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
2026-03-27 19:46:25 +01:00
David KayaandCopilot 1868a79d9a feat: add 'Always approve' button for tool-call approvals
When a tool-call approval banner appears, users can now click 'Always
approve' to both approve the current call AND add the tool to the
session's autoApprovedToolNames. Future calls to that tool are then
auto-approved without showing the banner.

- Add alwaysApprove?: boolean to ResolveSessionApprovalInput
- In resolveSessionApproval, atomically append tool to session
  approvalSettings when alwaysApprove is true (avoids running guard)
- Add 'Always approve' button in ApprovalBanner (tool-call kind only)
- Wire through ChatPane and App.tsx with updated callback signatures

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
2026-03-27 19:34:50 +01:00
David KayaandCopilot 4f1ae86021 fix: add diagnostic logging to MCP OAuth probing
Add [aryx oauth] log messages throughout the proactive auth flow so
failures are visible in the Electron main process console (DevTools →
Main Process or terminal output). Logs cover: probe start, HTTP status,
PRM discovery, token skip, flow start, and success/failure.

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
2026-03-27 19:31:03 +01:00
David KayaandCopilot f0c2b4982b feat: proactive OAuth when enabling HTTP MCP servers
When a user enables an HTTP MCP server for a session, proactively probe
the server URL. If it returns 401 and has discoverable OAuth metadata
(RFC 9728), automatically trigger the full OAuth 2.1 + PKCE flow and
open the browser for consent — matching VS Code's behavior.

- Add requiresOAuth() probe: GET server URL → check 401 → check PRM
- Add probeAndAuthenticateHttpMcpServers() in AryxAppService
- Call proactive probe from updateSessionTooling (fire-and-forget)
- Skip servers that already have stored tokens

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
2026-03-27 19:26:59 +01:00
David KayaandCopilot ac48aa58e0 feat: implement MCP OAuth 2.1 flow, token storage, and injection
- Create mcpTokenStore: in-memory token storage keyed by normalized server URL
  with automatic expiry checking
- Create mcpOAuthService: full OAuth 2.1 + PKCE flow implementation
  - Protected Resource Metadata discovery (RFC 9728)
  - Authorization Server Metadata fetch (RFC 8414)
  - Dynamic Client Registration (RFC 7591) when no static client ID
  - PKCE S256 code challenge generation
  - Local HTTP callback server for auth code receipt
  - Browser-based consent via Electron shell.openExternal
  - Authorization code to token exchange
- Add startSessionMcpAuth IPC channel and handler to trigger OAuth flow
- Inject stored OAuth tokens as Authorization headers in buildRunTurnToolingConfig
- Update McpAuthBanner with 'Authenticate in browser' button and loading state
- Add tests for token store (7 tests) and token injection (3 tests)

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
2026-03-27 19:15:37 +01:00
David KayaandCopilot f9757d5ce2 feat: add MCP OAuth authentication frontend wiring and banner
- Add McpOauthRequiredEvent and McpOauthStaticClientConfigEvent to sidecar contracts
- Add PendingMcpAuthRecord domain type with status tracking (pending/authenticating/failed/done)
- Add pendingMcpAuth field to SessionRecord for session-level auth state
- Wire onMcpOAuthRequired callback through sidecarProcess, runTurnPending, and AryxAppService
- Handle mcp-oauth-required events: set session pendingMcpAuth, clear on turn finalize/cancel
- Add dismissSessionMcpAuth IPC channel with preload binding and handler registration
- Create McpAuthBanner component (amber-themed, shows server name/URL, dismiss button)
- Integrate McpAuthBanner into ChatPane with placeholder text and App.tsx callback
- Update test fixtures for new RunTurnPendingCommand.onMcpOAuthRequired field

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
2026-03-27 19:09:11 +01:00
David KayaandCopilot 192c28f721 feat: surface MCP OAuth sidecar events
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
2026-03-27 19:01:46 +01:00
David KayaandCopilot b670680a7d refactor: improve plan mode UX with composer-integrated toggle and guidance text
- Move plan mode toggle from pills row to composer send button area
- Plan toggle sits next to the send button as a mode switch icon
- Send button turns emerald when plan mode is active
- Add mode indicator strip below composer when plan mode is on
- Replace 'Implement this plan' auto-send button with instructional
  guidance text that tells the user to send a follow-up message
- Keep Dismiss button to clear the plan review banner

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
2026-03-26 23:46:04 +01:00
David KayaandCopilot 231be36e6c feat: add plan mode frontend support with mode toggle and plan review UI
- Add InteractionMode type and ExitPlanModeRequestedEvent to sidecar contracts
- Add PendingPlanReviewRecord domain type and interactionMode to SessionRecord
- Wire onExitPlanMode callback through SidecarClient and RunTurnPendingCommand
- Pass session interaction mode to RunTurnCommand for sidecar consumption
- Handle exit-plan-mode-requested events in AryxAppService
- Add setSessionInteractionMode and dismissSessionPlanReview IPC methods
- Create PlanReviewBanner component with summary, markdown content, and actions
- Add plan mode toggle pill in ChatPane composer area
- Wire implement action as follow-up message (graceful degradation)
- Clear pending plan review on new turn, turn completion, and cancellation

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
2026-03-26 23:39:48 +01:00
David KayaandCopilot 380e402512 feat: add backend plan mode checkpoints
Add backend support for plan-mode turn shaping and exit-plan review
checkpoints. Run-turn commands now accept an interaction mode,
plan-mode prompt guidance tells agents to produce a plan and call
exit_plan_mode, and the sidecar emits a structured
exit-plan-mode-requested event when the SDK raises that session event.

For now the backend intentionally treats exit_plan_mode as a turn
boundary and graceful-degradation review checkpoint. The current
Agent Framework GitHubCopilotAgent wrapper does not expose a clean way
to bridge the live CopilotSession back into same-turn exit-plan
resolution, so that follow-up remains documented for the frontend and a
future backend slice.

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
2026-03-26 23:29:32 +01:00
David KayaandCopilot c069b86add docs: add website update guideline to AGENTS.md
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
2026-03-26 23:24:37 +01:00
David KayaandCopilot 912677fba0 feat: add tool metadata registry with human-readable labels
Add a builtin tool label registry mapping all ~30 official SDK tool IDs
to human-readable display names. Update listApprovalToolDefinitions to
resolve labels from the registry instead of showing raw tool IDs like
read_bash or fetch_copilot_cli_documentation. Expand the fallback
runtime tool catalog from 6 to 23 non-internal standard tools.

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
2026-03-26 19:27:36 +01:00
David KayaandCopilot a1a044e7ca fix: filter internal runtime tools
Exclude internal runtime tools from sidecar capability discovery so they do not surface in the approval tool catalog.

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
2026-03-26 19:23:48 +01:00
David KayaandCopilot f15b1aedb1 feat: add ask_user interactive user input to frontend
Wire the sidecar user-input-requested protocol event through the main
process, IPC layer, and renderer UI so agents can ask the user
interactive questions with choices and freeform input.

- Add UserInputRequestedEvent and ResolveUserInputCommand to sidecar
  protocol types
- Add resolveUserInput method to SidecarClient and onUserInput callback
  to runTurn
- Create PendingUserInputRecord domain type and add pendingUserInput to
  SessionRecord
- Add handleUserInputRequested and resolveSessionUserInput to
  AryxAppService with handle management
- Register sessions:resolve-user-input IPC channel with preload bridge
- Create UserInputBanner component with choice buttons and freeform
  input
- Integrate UserInputBanner into ChatPane with header indicator

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
2026-03-26 17:45:27 +01:00
David KayaandCopilot 2ae4bd01b4 feat: add ask_user support to sidecar
Implement the Copilot SDK user input round-trip in the backend:
- add user-input-requested and resolve-user-input protocol DTOs
- add a CopilotUserInputCoordinator that mirrors approval flow with
  pending TaskCompletionSource state and explicit resolution
- wire SessionConfig.OnUserInputRequest through CopilotAgentBundle
  and CopilotWorkflowRunner
- extend SidecarProtocolHost to emit user input events, accept
  resolve-user-input commands, and filter ask_user from runtime
  approval tools
- add regression tests for the new coordinator and protocol flow

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
2026-03-26 17:34:49 +01:00
David KayaandCopilot 4c01d8b1a7 feat: surface structured permission details in approval UI
Wire the sidecar's new permissionDetail payload through the frontend:
- Add PermissionDetail interface to sidecar contract
- Add permissionDetail field to PendingApprovalRecord and thread
  through normalization and AryxAppService mapping
- Create PermissionDetailView component with kind-specific rendering:
  shell (command block + warning), write (filename + diff), read (path),
  mcp (server badge + args), url (prominent URL), memory (subject/fact),
  custom-tool (description + args), hook (message + args)
- ApprovalBanner shows structured detail when available, falls back to
  generic text for backward compatibility
- QueuedApprovalsList shows brief kind-specific summaries

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
2026-03-26 16:59:15 +01:00
David KayaandCopilot 13c543e2fd feat: add structured tool approval details to sidecar protocol
Expose SDK permission request details on approval-requested events via a
new permissionDetail payload. This preserves the existing title/detail
summary while carrying the command, URL, diff, args, and other
kind-specific data needed for richer approval UI in the renderer.

Also adds sidecar coverage for all supported permission request kinds and
verifies the new payload serializes over the protocol.

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
2026-03-26 16:40:41 +01:00
David KayaandCopilot 69562c19f3 fix: fix Inno Setup preprocessor defines via environment variables
ISCC's /D command-line flag combined with Node.js spawn on Windows
produces broken ISPP defines due to argument quoting conflicts.
Switch to environment variables read via GetEnv() — robust and
avoids all quoting issues. Product name and publisher are now
hardcoded in the ISS (they never change).

Verified: installer shows 'Aryx' and correct version in metadata.

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
2026-03-26 16:14:14 +01:00
David KayaandCopilot 7c1852f79f refactor: replace NSIS with Inno Setup for Windows installer
Switch from NSIS (early 2000s) to Inno Setup 6 — the modern standard
used by VS Code and most Electron apps. Features:

- Modern wizard UI (WizardStyle=modern)
- LZMA2/ultra64 solid compression (138.9 MB, down from 143 MB with NSIS)
- Per-user install to %LOCALAPPDATA%\Programs\Aryx (no admin needed)
- Optional desktop shortcut, Start Menu group
- Auto-closes running instance before install/uninstall
- Proper Add/Remove Programs integration via AppId
- Launch-after-install option

CI updated to install Inno Setup via Chocolatey on windows-latest.

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
2026-03-26 14:47:11 +01:00
David KayaandCopilot d7d1b33a53 fix: fix CI installer creation for all platforms
- Windows: add NSIS installation step via Chocolatey in CI workflow
- macOS: fix create-dmg CLI invocation to match sindresorhus package
  API (--overwrite, --no-version-in-filename, --no-code-sign) and
  rename output to expected asset name
- Add rename import to create-installer.ts

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
2026-03-26 13:35:20 +01:00
David KayaandCopilot f838d2ac15 feat: add native platform installers for Windows, macOS, and Linux
Replace zip/tar archive distribution with platform-native installers:

- Windows: NSIS installer (.exe) with per-user install to
  %LOCALAPPDATA%\Programs\Aryx, Start Menu and Desktop shortcuts,
  Add/Remove Programs registration, and silent install support
- macOS: DMG disk image (.dmg) with drag-to-Applications experience
  via create-dmg
- Linux: Debian package (.deb) with /opt/aryx installation,
  /usr/bin/aryx symlink, desktop entry, hicolor icons, and
  libsecret-1-0 dependency declaration

Add scripts/create-installer.ts as the platform-dispatching entry
point. Update CI workflow to produce native installers instead of
archives. Add installerAssetName to ReleaseTarget interface.

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
2026-03-26 13:06:50 +01:00
David KayaandCopilot da7aa5cdd2 perf: enable .NET sidecar trimming and single-file publish
- Enable PublishTrimmed with TrimMode=partial to remove unused
  framework assemblies while preserving app/SDK code safety
- Enable PublishSingleFile to bundle all managed assemblies into
  one executable (sidecar goes from 254 files to 2 files)

Sidecar shrinks from 191 MB / 254 files to 134 MB / 2 files.

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
2026-03-26 12:22:30 +01:00
David KayaandCopilot d1cc3c2e6d perf: optimize release artifact size and file count
- Move 15 renderer-only dependencies to devDependencies since they are
  already bundled by Vite (197 → 37 runtime deps)
- Pack app resources into asar archive with native addon unpacking
- Strip unused Electron locale files, keeping only en-US
- Remove non-essential Electron files (LICENSES.chromium.html, LICENSE)
- Remove default_app.asar from Electron distribution
- Strip .NET sidecar localization satellite assemblies (en only)

Release artifact shrinks from 607 MB / 10,525 files to 478 MB / 235 files.

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
2026-03-26 12:19:46 +01:00
David KayaandCopilot cb05148231 fix: require child_process in gitService
Use createRequire for node:child_process so Bun on macOS can load gitService without ESM export-shape issues.

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
2026-03-26 00:37:24 +01:00
David KayaandCopilot 66c129f8b9 fix: handle bun child_process interop
Resolve execFile from named or default child_process exports so macOS Bun can load gitService in tests.

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
2026-03-26 00:34:44 +01:00
David KayaandCopilot d9f3f5302a fix: resolve macos bun interop
Use Bun-safe Electron and child_process import patterns for macOS GitHub Actions tests.

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
2026-03-26 00:30:41 +01:00
David Kaya 40ef1c07b5 fix: failing macos test 2026-03-26 00:22:28 +01:00
David KayaandCopilot 88030eb48d refactor: finish Aryx rebrand sweep
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
2026-03-26 00:17:14 +01:00
David KayaandCopilot 1edd3aed55 refactor: rename sidecar host to Aryx
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
2026-03-26 00:14:12 +01:00
David KayaandCopilot 4e3c74497f feat: add macOS arm64 release build
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
2026-03-26 00:06:07 +01:00
David KayaandCopilot 90cc32e737 fix: update macOS GitHub Actions runner
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
2026-03-26 00:04:25 +01:00
David Kaya d02ccf22a2 feat: website changes 2026-03-26 00:01:32 +01:00
David KayaandCopilot d88d00df0d fix: make copilot CLI path resolution cross-platform
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
2026-03-25 23:44:49 +01:00
David Kaya f94c3232b4 fix: screenshot 2026-03-25 23:36:50 +01:00
David KayaandCopilot 5ed2b8bcd4 feat: add product website for Cloudflare Pages
Single-page marketing site built with Astro 5 and Tailwind CSS v4.
Produces static HTML with zero client-side JS — ideal for Cloudflare
Pages deployment.

Sections: hero with screenshot placeholder, 6 feature cards,
5 orchestration patterns showcase, 4-step getting started guide,
tech banner, and footer.

Dark theme (zinc-950/indigo accents) matches the app aesthetic.
Responsive design with mobile navigation.

Build: cd website && bun install && bun run build
Output: website/dist/

To add a real screenshot, replace
website/public/images/screenshot-placeholder.svg with
website/public/images/screenshot.png and update the img src
in website/src/pages/index.astro.

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
2026-03-25 22:57:00 +01:00
David KayaandCopilot 968ae37279 feat: allow model selection in all single-agent sessions
Generalize the scratchpad-only model/reasoning-effort selector to work
in any session whose pattern has exactly one agent. This enables model
switching above the chat input for project 1-on-1 chats, not just
scratchpads.

- Rename ScratchpadSessionConfig -> SessionModelConfig across domain,
  contracts, IPC, preload, and renderer
- Replace isScratchpadProject guard with agents.length === 1 check in
  EryxAppService.updateSessionModelConfig and ChatPane pills gate
- Apply session model config in buildEffectivePattern whenever present,
  regardless of project type
- Seed sessionModelConfig on session creation for any single-agent
  pattern

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
2026-03-25 22:41:46 +01:00
David Kaya 46b5de472e tests: some additional tests 2026-03-25 22:28:36 +01:00
David KayaandCopilot 628fdff29e fix: clear live agent status on cancel
Clear only active agent statuses when a session reaches idle, and keep the terminal cancelled/error run cleanup in place. Add regression coverage for idle cleanup after cancellation-like flows.

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
2026-03-25 22:20:57 +01:00
David KayaandCopilot c9cc2fd8e7 fix: clear stale agent activity on cancel
Clear only still-active side-panel agent states when a run ends as cancelled or error, while preserving completed agent statuses. Add a renderer regression test for cancelled runs.

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
2026-03-25 22:17:33 +01:00
David KayaandCopilot 82fdadb312 feat: add turn cancellation UI and main process wiring
Adds cancel-turn IPC channel, SidecarClient.cancelTurn(), TurnCancelledError,
EryxAppService.cancelSessionTurn(), finalizeCancelledTurn(), stop button in
ChatPane, and cancelled run status throughout the run timeline.

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
2026-03-25 22:17:15 +01:00
David KayaandCopilot d84b3021f2 feat: add sidecar turn cancellation
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
2026-03-25 22:04:14 +01:00
David KayaandCopilot c8bb9d6f59 fix: align chat MCP tooling with session project
Resolve chat tooling from the selected session's project instead of the
separately tracked selectedProjectId so project-scoped discovered MCPs do
not appear for the wrong session. Also synchronize selectedProjectId when
a session is selected and add regression tests for both behaviors.

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
2026-03-25 21:57:30 +01:00
David KayaandCopilot e5878ba9e8 feat: auto-discover MCP servers from project and user config files
Scan .vscode/mcp.json, .mcp.json, .copilot/mcp.json (project-level) and
~/.copilot/mcp.json (user-level) for MCP server definitions. Discovered
servers require explicit user acceptance before activation.

Backend:
- Add ConfigScannerRegistry with per-format scanners and  substitution
- Add discovered tooling domain model with fingerprint-based change detection
- Integrate scanning into workspace load, project add, and project selection
- Merge accepted discovered MCPs into effective runtime tooling
- Extend sidecar contracts with env/headers for real-world MCP configs
- Add IPC channels for accept/dismiss/rescan operations

Frontend:
- Add DiscoveredToolingModal for reviewing pending MCP servers
- Auto-show modal when pending discoveries exist on load or project switch
- Add amber badge on sidebar project headers for pending discoveries
- Add discovered MCP section in Settings panel with accept/dismiss/rescan
- Group InlinePills tool dropdown by Workspace MCP / User MCP / Project MCP
- Pass effective project tooling (including accepted discoveries) to ChatPane

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
2026-03-25 21:54:06 +01:00
David KayaandCopilot 08a1132d86 refactor: inline tool auto-approval list into tool-call approval card
Move the Tool Auto-Approval Defaults section into the tool-call
ApprovalCheckpointRow card so the tool list appears directly under
the toggle instead of in a disconnected section below. Extend
ApprovalCheckpointRow with an optional children prop rendered
inside the card when the checkpoint is enabled.

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
2026-03-25 21:51:15 +01:00
David KayaandCopilot 2e3175934d feat: add troubleshooting settings with open app data and reset workspace
Add a Troubleshooting section under a Support nav group in Settings:

- Open App Data Folder: reveals the user-data directory in the OS file
  explorer via shell.openPath.
- Reset Local Workspace: destructive action that deletes workspace.json
  and scratchpad contents, clears the in-memory cache, and reseeds
  defaults. Auth secrets are preserved. Requires two-step confirmation.

New IPC channels, ElectronApi methods, and preload bridge added for
openAppDataFolder and resetLocalWorkspace. Existing electron mock in
tests updated to include shell.

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
2026-03-25 21:50:16 +01:00
David Kaya 5123f9163c docs: update agent guidelines 2026-03-25 21:41:04 +01:00
David Kaya 7957928fc7 fix: inline pills 2026-03-25 21:40:40 +01:00
David KayaandCopilot 29c054f47d fix: require tool approval by default
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
2026-03-25 21:23:01 +01:00
David KayaandCopilot cc3c04b841 feat: enable scratchpad tooling frontend and instruction update
- Remove !isScratchpad gate on tooling/approval pills in ChatPane
- Update scratchpad agent guidance to permit tool/file usage
- Add EryxAppService scratchpad tooling integration tests
- Update ARCHITECTURE.md to reflect scratchpad tooling support

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
2026-03-25 21:20:43 +01:00
David KayaandCopilot 4ecfa5bff6 feat: enable scratchpad tooling backend
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
2026-03-25 21:14:19 +01:00
David KayaandCopilot ab4e217d74 refactor: rename app to aryx
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
2026-03-25 21:11:36 +01:00
David Kaya cb7f216bc3 fix: styling 2026-03-25 20:43:04 +01:00
David KayaandCopilot 8d3c4f474e revert: replace plain CSS class with @theme text-pill token
Remove .pill-trigger CSS class and restore the idiomatic Tailwind v4
@theme approach: --text-pill in @theme generates the text-pill utility.

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
2026-03-25 20:40:11 +01:00
David KayaandCopilot 4797c9b30c fix: use plain CSS class for pill font size instead of Tailwind token
The @theme --text-pill token and text-[8px] arbitrary value both
failed to apply at dev time due to @tailwindcss/vite not generating
the utility rule during HMR. Use a plain .pill-trigger CSS class
in styles.css instead, which is always available regardless of
Tailwind's JIT scanner behavior.

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
2026-03-25 20:40:10 +01:00
David KayaandCopilot d6596efc11 fix: register pill font size as Tailwind v4 theme token
The arbitrary text-[8px] class was not being generated by the
Tailwind v4 Vite plugin at dev time. Register --text-pill in
@theme so text-pill becomes a first-class utility that is always
generated. Uses 0.5rem (8px) with line-height 1.

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
2026-03-25 20:33:28 +01:00