27 Commits

Author	SHA1	Message	Date
jijiamoer	5c0d4f5e9b	feat: add Devin ACP agent integration (#1497)	2026-07-02 21:57:25 +01:00
Carlos Florêncio	49c482bd4d	fix: prevent leaked agent processes on shutdown (#1498)	2026-06-26 10:09:23 +01:00
Carlos Florêncio	c724e38be9	fix: apply profile auto-approve to agentctl instances (#1358)	2026-06-13 23:22:54 +01:00
Carlos Florêncio	3e92d3abc6	fix: allow office mcp mode for existing tasks (#1315)	2026-06-08 21:22:30 +01:00
João Salavisa	d83c1dd982	feat(diff): use stored base_branch for diff stats + per-task compare picker (#1273) ... * feat(diff): use stored base_branch for diff stats + per-task compare picker Task-card additions/deletions and the Changes panel were computed against a hardcoded origin/main → master priority list inside agentctl. For tasks branched off a release/staging branch this inflated the counts by every commit since the actual fork point, and the Changes panel grouped unrelated files alongside the task's real work. This wires the per-repo task_repositories.base_branch end-to-end into the agentctl WorkspaceTracker so each repo's diff stats resolve against the recorded base instead of the integration-branch fallback. A picker on the Changes panel branch hover card lets the user change the base mid-task; the new value is persisted via PATCH /tasks/:id/repositories/:repo_id, pushed live to the running agentctl instance, and the session's cached base_commit_sha is cleared so the commits panel + cumulative diff refetch against the new base without a session restart. Falls back to the existing origin/main → master list for legacy tasks (empty base_branch) and to the branch tip when no merge-base exists with HEAD (unrelated histories — local rebase backups, etc.) so the stats and commits panels stay anchored on the same SHA in every failure mode. * fix(agentctl): sanitize user-controlled base_branch refs before git exec CodeQL flagged the picker's user-controlled base_branch payload reaching exec.Command("git", ..., baseBranch) inside workspace_git_cmd.go as a command-built-from-user-controlled-source vulnerability. A malicious caller could smuggle a leading-"-" argument like "-upload-pack=..." that git itself reinterprets as a flag, even though exec.Command keeps args separate from the shell. Adds an IsSafeGitRef sanitizer at every boundary the value crosses on its way to git: the service write path, the agentctl HTTP handlers (/workspace/base-branches, /git/log, /git/cumulative-diff), and WorkspaceTracker.SetBaseBranch itself. Unsafe values are rejected at the service tier and silently downgraded to the no-override fallback inside agentctl so a misbehaving caller can't half-break the tracker. * fix(diff): address PR review blockers and CI lint - Split UpdateRepositoryBaseBranch into validate/load/apply helpers to drop cyclomatic complexity below the cyclop limit that gated Run Backend Tests. - Replace direct m.cfg.BaseBranches reads/writes with getBaseBranches / setBaseBranches under a dedicated RWMutex. UpdateBaseBranches used to write under repoTrackersMu while rescan + lazy-subpath callers read under workspaceTrackersMu, leaving an unsynchronized data race on the map. - Drop the false 'PR target is auto-wired' claim from the MCP tool description — GitOperator.CreatePR takes baseBranch as an explicit arg and never reads the tracker's value. - Picker no-op guard now compares against currentBase instead of storedBase so selecting the displayed fallback (legacy task before any picker change) is treated as no-op instead of triggering a spurious save. - HTTP handler now maps non-validation service errors to 500 instead of 400, classifying via the existing isValidationError helper plus the new unsafe-git-ref message. * fix(diff): regex-based ref sanitiser at every git subprocess sink CodeQL kept flagging workspace_git_cmd.go:43 even after SetBaseBranch rejected unsafe values — the taint tracker doesn't follow sanitised stores through struct fields, so it still saw user input flowing into exec.Command("git", ..., baseBranch). Switches IsSafeGitRef to a single regex check (a pattern CodeQL recognises as a sanitiser barrier) and adds SanitizeGitRef calls inline at every agentctl call site that splices a user-controlled ref into a git arg: - workspace_git_status.resolveStoredRef (before the rev-parse probes) - workspace_git_status.computeBaseCommit (before merge-base/rev-parse) - workspace_git_status.getAheadBehindCounts (before rev-list) - agentctl/server/api/git.go runGitLogForRepo + handleGitCumulativeDiff (before computeMergeBase/GetRevParse on the HTTP TargetBranch) Behaviour is unchanged for valid refs — only narrows the allowlist (now [A-Za-z0-9_./-]) so the sanitiser barrier is recognisable to static analysis. SetBaseBranch's field-store sanitiser stays as defence in depth. * fix(diff): transform sanitiser at every map read for CodeQL taint flow CodeQL still flagged workspace_git_cmd.go:43 because: - handleSetBaseBranches used an IsSafeGitRef bool guard before a verbatim map[k] = v copy. CodeQL recognises regex-transforming sanitisers but not bool-guarded passthroughs. - lookupBaseBranch returned m.cfg.BaseBranches[k] verbatim. Even though the map was sanitised on write, taint flow across map stores + field stores loses the sanitised state in the analyser. Switches both sites to the transforming SanitizeGitRef helper so every read produces a freshly-sanitised value before it flows to a downstream git argument. Behaviour is unchanged for valid refs. * fix(diff): regex transformer sanitiser + guard live push from empty map CodeQL still flagged the previous SanitizeGitRef because it returned the input string unchanged on success — an identity passthrough that the taint tracker doesn't recognise as a sanitiser barrier. Switches to safeGitRefPattern.FindString so the output comes from the regex match (not the parameter). IsSafeGitRef now wraps SanitizeGitRef so the predicate and the transformer share one allowlist. Also addresses two Claude review blockers: - collectTaskBaseBranches returns a nil/empty map when the task has no recorded base_branches; pushing that to agentctl wipes every tracker's override. Now skipped — the DB row is the source of truth for the next session launch. - applyBaseBranchSideEffects now runs on context.WithoutCancel so an HTTP / WS request cancel after the task_repositories UPDATE commits can't leave session.base_commit_sha stale + the live agentctl tracker map un-pushed, drifting from the persisted row. * fix(diff): share securityutil.IsValidBranchName + harden error classification CodeQL still flagged the picker's regex sanitiser as taint passthrough. Refactors SanitizeGitRef / isSafeBaseBranchRef to delegate to the existing securityutil.IsValidBranchName helper that agentctl's Rebase / Merge / RenameBranch already use without being flagged. The shared allowlist also fixes a Claude review blocker: 'feature/@2024' used to pass the service validator but get dropped by the agentctl-side check at push time, silently wiping every tracker's override. Plus the two error-classification blockers Claude raised on the previous HEAD: - WS handler (task_ws_handlers.go wsUpdateTaskRepository) now distinguishes validation errors (surfaced verbatim) from internal faults (opaque 500-style message). HTTP path already did this via the existing isValidationError helper. - MCP handler (handlers.handleUpdateRepositoryBaseBranch) gets the same treatment via a new isValidationError local to the mcp/handlers package so raw service error strings stop leaking across the MCP boundary. * fix(diff): inline securityutil.IsValidBranchName at every git subprocess sink CodeQL's interprocedural taint tracking wasn't recognising SanitizeGitRef as a sanitiser even after it delegated to securityutil.IsValidBranchName — the wrapper helper + map/field hops between source and sink kept the flow tainted in the analyser. Inlines the IsValidBranchName allowlist check (already used inline by Rebase/Merge in git.go without being flagged) at every call site that splices a user-controlled ref into a git subprocess argument: - workspace_git_status.resolveStoredRef (before rev-parse --verify) - workspace_git_status.computeBaseCommit (before merge-base / rev-parse) - workspace_git_status.getAheadBehindCounts (before rev-list) - agentctl/server/api/git.go runGitLogForRepo + handleGitCumulativeDiff - agentctl/server/api/workspace_base_branches.go handleSetBaseBranches Behaviour unchanged: same allowlist as Rebase / Merge / RenameBranch already use. SanitizeGitRef / IsSafeGitRef survive as defence-in-depth helpers for callers that want a one-line check; they delegate to the same securityutil helper so the rules stay aligned. * fix(diff): inline regexp.MatchString at every git subprocess sink CodeQL still flagged workspace_git_cmd.go:43 even with the inline securityutil.IsValidBranchName helper — the taint tracker apparently does not follow into helper functions for the go/command-injection rule on this path. Hoists the regex to package-scope vars (safeBranchRefPattern in both process and api packages) and calls regexp.MatchString directly at each sink site. Behaviour is identical to securityutil.IsValidBranchName; only the call form changes. * fix(diff): SHA-shape allowlist on BaseCommit before downstream git diff CodeQL still flagged workspace_git_cmd.go:43 because update.BaseCommit is the OUTPUT of an earlier merge-base / rev-parse call whose argument list contained a user-controlled branch ref — the analyser treats that command's output as still tainted even when the input ref was sanitised. Adds an inline 40-hex SHA-shape allowlist check in enrichWithBranchDiff right before `git diff --numstat <update.BaseCommit>` so the sanitiser barrier sits in the same function as the next subprocess call. Behaviour is unchanged: merge-base / rev-parse already produce SHA-shaped output. * fix(diff): async tracker refresh + tighten init spread order Two CodeRabbit/Claude review fixes: - UpdateBaseBranches stamps each tracker's new baseBranch synchronously then runs RefreshGitStatus on a background goroutine instead of inside the picker-save HTTP handler. Per-tracker git subprocesses can each take ~3-5 calls; serialising them on the request goroutine made the PATCH response wait for every refresh to finish. The picker UI re-fetches via the existing WS stream once Refresh emits, so the handler doesn't need to block. - updateTaskRepositoryBaseBranch now spreads options.init BEFORE the required method/body fields so an upstream caller's init overrides can't replace the PATCH method or the base_branch payload. Matches the convention deleteTask / moveTask already follow. * fix(diff): narrow mcp isValidationError + opaque 500 + post-sanitize log count Three final Claude review threads on the latest commit: - mcp/handlers isValidationError drops the broad 'invalid' substring so DB / IO errors whose messages happen to contain 'invalid' don't misclassify as ErrorCodeValidation. - task_http_handlers httpUpdateTaskRepository now logs the raw service error server-side and returns an opaque 'failed to update task repository' on the 500 path. Stops potential DB / IO detail leakage through the HTTP boundary. - workspace_base_branches debug log reports accepted + rejected counts (post-sanitisation) instead of the pre-filter total so the metric actually reflects what hit UpdateBaseBranches. --------- Co-authored-by: Kandev Agent <agent@kandev.dev>	2026-06-05 15:53:39 +01:00
João Salavisa	6a684211fc	fix(agentctl): reap disconnected ACP sessions + MCP child trees (#1249) ... * fix(agentctl): reap disconnected ACP sessions + MCP child trees (#1247) Two independent leaks let opencode acp sessions accumulate indefinitely: 1. instance.Manager never reaped disconnected sessions. StopInstance only ran via the kandev DELETE route, so any session whose backend went away (UI closed mid-run, WS dropped, opencode failed to emit complete) was orphaned forever. 2. Even when DELETE fired, the ACP adapter unconditionally returned RequiresProcessKill=false. opencode acp keeps an HTTP server + MCP child tree alive after stdin EOF, so closing stdin was a no-op and the children re-parented to init. Fix both: - Plumb RuntimeConfig.RequiresProcessKill from agents.OpenCodeACP through the ExecutorCreateRequest -> CreateInstanceRequest -> InstanceConfig -> adapter.Config -> shared.Config chain. acp.Adapter now reads cfg.RequiresProcessKill instead of hardcoding false. - Belt-and-braces: process.Manager.waitForProcessExit timeout fallback now killProcessGroup first (only falls back to single-process kill if that errors), so any future ACP CLI that ignores stdin EOF still gets reaped. - Add an idle-instance reaper to instance.Manager. Per-instance activity middleware tracks inflight HTTP requests + lastActivity; a background goroutine sweeps every minute and StopInstance's any instance idle past KANDEV_ACP_IDLE_TIMEOUT (default 1h, set to 0 to disable). Long-lived /agent/stream WS connections keep the inflight counter above zero so active sessions are never reaped. - Validate stdio MCP commands at CreateInstance time via exec.LookPath; missing entries are dropped with a warn instead of spawning a permanently broken child every session (fixes the stale /snap/bin/brave MCP repro in the issue). Regression tests: process-group kill (parent+child fixture), idle reaper end-to-end, MCP validation, opencode-acp opt-in vs other ACP agents' default. * fix: address PR review feedback on #1247 fix - idle_reaper: extract reapSingleIdleInstance helper with defer cancel(), add reaperStop polling between sweep iterations so Shutdown drains in one StopInstance window instead of N * timeout (cubic, claude, greptile). - instance.Manager: use sync.Once instead of m.mu to gate reaper-stop channel close (claude). - mcpStdioValidationError: tolerate compound Command strings by resolving only the first whitespace-delimited token; reject entries with neither URL nor Command outright; document agentctl-vs-agent PATH skew in Docker/SSH executor modes (claude, greptile, cubic). - manager_killgroup_test: treat zombie state as dead via /proc/<pid>/stat on Linux + bump Eventually window to 15s — CI containers' PID 1 can delay harvesting orphans long enough to flake signal(0) probes. - instance package: add goleak TestMain for the new reaper goroutine (coderabbit). - AGENTS.md: correct SIGKILL vs SIGTERM wording; clarify the timeout fallback only fires on ctx.Done (coderabbit, cubic). --------- Co-authored-by: Kandev Agent <agent@kandev.dev>	2026-06-02 21:46:49 +01:00
José Almeida	b109c84e4e	fix(backend): pass HTTP MCP servers to ACP agents via AssumeMcpHttp (#1152) ... * fix(backend): pass HTTP MCP servers to ACP agents via AssumeMcpHttp * test(backend): cover AssumeMcpHttp override path with effectiveMcpCapabilities helper	2026-05-29 12:58:51 +01:00
João Salavisa	f98395f000	fix(agentctl): start per-repo workspace trackers in passthrough mode (#1002) ... * fix(agentctl): start per-repo workspace trackers in passthrough mode CLI passthrough mode starts the agent CLI via the PTY-based InteractiveRunner instead of agentctl's startAgent / startOneShot, so the per-repo workspace trackers created for multi-repo task roots were never started. The bare root tracker exits immediately because it has no git index, leaving polling completely idle — file edits made by the agent never produced workspace-stream notifications and the Files panel stayed stale. Move the tracker startup to instance.Manager.CreateInstance via a new StartAllWorkspaceTrackers helper so polling fires for every repo regardless of whether the agent runs via ACP or passthrough. Fixes #982 Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(lifecycle): serialize workspace poll-mode pushes to prevent slow-stuck mode When a session subscribes and then immediately focuses (the normal page-load sequence), the workspace poll aggregator used to spawn two independent goroutines — one pushing slow, one pushing fast — whose HTTP arrival order at agentctl was undefined. With the slow request landing after the fast request the tracker ended up in slow mode (~30s polling) while the aggregator's lastPushed claimed fast, and the suppress-duplicate guard then blocked any future focus event from reconciling the divergence. Multi-repo passthrough sessions hit this every page load: the Files panel saw ~30s-stale state even with the per-repo trackers running. Replace the fire-and-forget goroutine per push with a single in-flight pusher per workspace plus a last-write-wins pending slot — concurrent callers all converge on the same pusher, which drains the queue in order so the final queued mode is what reaches the wire last. Regression test in manager_subscription_test.go simulates the race with an httptest server that adds artificial latency on slow pushes; the test fails on the old parallel-goroutine implementation and passes on the serialized one. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(agentctl): address PR review feedback for passthrough tracker fix - Goroutine leak: extract stopWorkspaceTrackers() and call it unconditionally at the top of Manager.Stop(). Previously the early return when status==Stopped skipped tracker teardown in passthrough mode (where Start() never runs), leaking 2 goroutines per per-repo tracker per stopped instance. stopShellAndProcesses delegates to the same helper. - Trim multi-line doc comments to one-liners per CLAUDE.md (the StartAllWorkspaceTrackers docstring and the CreateInstance call-site comment). - Replace the fixed time.Sleep in the race regression test with a poll on len(calls), exiting as soon as the pusher has drained, plus a brief settle window after to keep the test sensitive to the late racing slow push the fix guards against. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(web): refresh expanded subtrees on polling-detected file changes The agentctl workspace tracker emits a generic `refresh` event with empty path when its poll cycle detects file changes (it knows something changed but not which file). The Files panel previously only refreshed the root folder on these events, so files added under an expanded subdir never became visible — the merge preserved the stale subtree, and the user had to manually collapse + re-expand to see new files. This is the user-visible symptom of #982 (Files panel does not refresh after agent makes changes in CLI passthrough mode): the backend fixes in this PR get the events flowing, but the frontend needs to act on them correctly. On a `refresh` event we now refresh every currently-expanded folder in addition to the root; for multi-repo task roots we scope to the affected repo's branch using `repository_name`. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix: address iteration-2 review feedback - Trim doc + inline comment blocks per CLAUDE.md (manager.go, file-browser-hooks.ts, workspace-files.ts). - Replace the hand-rolled fmtSscanf/numErr port parser with strconv.Atoi in the race regression test. - Assert on slowCalls in the race test to lock in the queue-collapse guarantee (at most one slow push lands). Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix: address iteration-3 review feedback - Remove redundant stopWorkspaceTrackers call from stopShellAndProcesses (it's already invoked at the top of Manager.Stop), and drop the now- misleading "stopping workspace tracker" debug log from the regular shutdown path. - Trim three multi-line comment blocks in manager_subscription.go to one-liners per CLAUDE.md (workspacePushTarget, pushAsync, pushLoop). Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix: trim multi-line comment block on pendingPush field * fix: trim test docstring on race regression test * fix(web): preserve sibling-repo subtrees on scoped refresh The setTree reducer ran patchNode(updated) at the root, which re-matched folderUpdates.has("") and overwrote the already-merged root children with the raw depth-1 server response — wiping the loaded children of every top-level repo that wasn't part of the refresh scope. Recurse over updated.children instead of updated itself: the root has already been merged in the block above, and each top-level repo either gets its targeted update (if folderUpdates.has(path)) or is left alone (preserving its loaded subtree). Includes a regression test that fails on the old code: a refresh scoped to "thm" used to discard "kandev"'s loaded README.md child. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(web): trim three more comment blocks to one-liners * fix: trim five more comment blocks to one-liners --------- Co-authored-by: Kandev Agent <agent@kandev.dev> Co-authored-by: Claude Opus 4.7 <noreply@anthropic.com>	2026-05-22 01:41:56 +01:00
Carlos Florêncio	0b2085688b	feat(office): autonomous agent management layer (#914)	2026-05-16 20:07:57 +01:00
Carlos Florêncio	684ea1f98e	fix(mcp): carry env and headers through MCP server config pipeline (#439)	2026-03-14 21:59:41 +00:00
Carlos Florêncio	0c77056304	fix: server-side task ID injection for plan tools and UI polish (#431)	2026-03-13 11:53:42 +00:00
José Almeida	f1da9644f4	feat: agent-native config mode (#396) ... * feat: add MCP mode pipeline for config vs task tool registration Thread McpMode field through the entire instance creation chain: CreateInstanceRequest → InstanceOverrides → InstanceConfig → MCP Server. This allows config-mode MCP sessions to register a different set of tools than the default task-mode sessions. * feat: add config-mode MCP tools and agentctl handlers Register config-mode tools (workflow steps, agents, MCP config, tasks) when MCP server mode is "config". Add 16 new WS action constants and agentctl-side handlers that forward tool calls to the backend via the existing BackendClient channel. * feat: add backend MCP handlers for config-mode operations Add backend WebSocket handlers for workflow step CRUD, agent CRUD, MCP config get/update, and task move/delete/archive/state operations. Wire config dependencies via SetConfigDeps pattern to avoid breaking the existing NewHandlers constructor. * feat: add config-mode system prompt and unit tests Add ConfigContext system prompt listing all config-mode MCP tools. Add unit tests for mode-based tool registration, agentctl config handler forwarding, backend handler validation, and system prompt content. * test: add exhaustive E2E tests for config-mode MCP handlers 38 end-to-end tests covering the full handler → service → DB pipeline for all 16 config-mode operations: workflow step CRUD, agent CRUD, agent profiles, MCP config, and task management (move, delete, archive, update state). Includes handler registration verification and two full pipeline integration tests. * fix: remove E2E test that depends on agent CLIs installed in CI TestE2E_Agent_CreateMultipleAndList tried to create claude-code and codex agents, which aren't available in CI. Agents are unique by type ID so creating duplicates also fails. Removed the test since CreateAndList already validates the create+list flow. * test: remove Go E2E tests for config-mode handlers Replacing with Playwright E2E tests. * test: add Playwright E2E tests for config-mode features 13 tests covering workflow steps, task operations, agent/profile settings pages, MCP config, and workflow settings round-trips. * feat: add dynamic MCP mode reconfiguration for agentctl instances Add SetMode() to the MCP server that clears and re-registers tools for the new mode. Expose it via PUT /api/v1/mcp/mode on per-instance servers, a SetMcpMode client method, and a lifecycle Manager.SetMcpMode wrapper. This enables switching from task to config MCP tools on workspaces that were prepared before plan_mode was set on the session. * feat: wire McpMode through executor launch and resume pipelines Add McpMode to LaunchOptions so callers can set it at launch time. StartTask and StartTaskWithSession now pass McpMode="config" when plan mode is active. startAgentOnExistingWorkspace reconfigures the MCP server via SetMcpMode before starting the agent subprocess. The AgentManagerClient interface gains SetMcpMode; the lifecycle adapter, executor, and all mock implementations are updated. * fix: pass plan_mode from workflow step when auto-starting moved tasks handleTaskMovedNoSession was hardcoding planMode=false when calling StartTask. Now it loads the target step and checks for enable_plan_mode in on_enter actions, passing the correct value so config-mode MCP tools are registered for the agent. * test: fix config-mode E2E tests and add SetMcpMode mock stubs - Add SetMcpMode stubs to all AgentManager mock implementations - Fix strict mode violations with { exact: true } on tool name assertions - Expand collapsed tool call groups before asserting individual names - Remove temporary diagnostic code * fix: address CodeRabbit review comments - Add servers parameter to update_mcp_config tool schema - Add mutex protection to SetMode for thread safety - Create session before building prompt so config context has session ID - Fix inconsistent ask_user_question tool name reference (use _kandev suffix) * refactor: use canonical ws.ActionMCP constants and handle marshal errors - Import ActionMCP* constants from pkg/websocket instead of duplicating them locally in config_handlers.go and handlers.go - Handle json.MarshalIndent error in forwardToBackend instead of ignoring it * fix: extract McpModeConfig constant and fix Prettier formatting - Add McpModeConfig constant to avoid repeated "config" string (goconst) - Fix Prettier formatting in config-management E2E spec * fix: validate task state before update and strengthen sysprompt tests - Add TaskState validation in handleUpdateTaskState to return proper validation error instead of internal error for invalid states - Strengthen format-contract assertions (exact %s counts, no %! markers) - Extract repeated test data into constants per project guidelines * refactor: decouple config MCP mode from plan_mode, use session metadata Replace plan_mode-based config MCP activation with config_mode in session metadata. Add isConfigModeSession() helper used across executor_execute, executor_interaction, and executor_resume. Propagate task.Metadata to session.Metadata in PrepareSession so config_mode flows from task creation through to the agent launch. * refactor: remove task tools from config MCP mode and update sysprompt Remove task management tools (list_tasks, move_task, delete_task, archive_task) from ModeConfig registration — they remain available in ModeTask. Update ConfigContext system prompt to remove task tool docs and add example requests section. Update MCP server tests for new tool count (18→14) and remove deleted task handler tests. * feat: add config-chat endpoint and workspace config agent profile Add default_config_agent_profile_id column to workspaces with SQLite migration. Add POST /workspaces/:id/config-chat endpoint that creates an ephemeral task with config_mode metadata and starts a session with config MCP tools. Extract resolveConfigChatDefaults helper to stay within lint complexity limits. * feat: add config chat UI with workspace settings integration Add config chat modal component and hook for starting config sessions. Add configChat state slice with open/close actions. Add workspace API client for config-chat endpoint. Add configuration agent profile selector to workspace settings form. Add config chat button to settings layout sidebar. * test: rewrite config-mode E2E tests to use metadata and session_id Rewrite config-management E2E tests to use createTaskWithAgent with config_mode metadata instead of a custom fetch helper. Fix navigation to use task.session_id instead of task.id. Add metadata parameter support to ApiClient.createTaskWithAgent. * fix: use floating FAB for config chat and pass default agent profile Move config chat button from header to a floating action button in the bottom-right corner with a sparkles icon. Pass the workspace's default agent profile ID when opening the config chat to avoid "no agent profile configured" error. Add default_config_agent_profile_id to workspace slice type. * feat: async agent launch for config-chat and extract helpers The config-chat endpoint now launches the agent asynchronously when a prompt is provided, matching the start_agent flow used by task creation. Extract deleteTaskOnError and launchConfigChatAgent helpers. Remove LaunchWorkspace from prepare step to avoid blocking. * feat: first-interaction agent profile selection for config chat Open the config chat modal immediately on FAB click. If no default config agent profile is set, the empty state now shows a profile selector before starting. The selected profile is saved as the workspace default_config_agent_profile_id for future sessions. * test: fix config E2E tests — remove duplicate method, use real tools Remove duplicate startConfigChat method in ApiClient that overrode the correct 3-param signature. Replace invalid create_agent test (requires registered agent type) with update_agent test. Remove fragile tool calls group UI assertion from multi-tool test. * fix: resolve 'no active session' error in config chat Add openConfigChatModal action that only sets isOpen and workspaceId without touching sessionId. Previously, open() called openConfigChat with empty sessionId which was truthy, causing QuickChatContent to render with an invalid session. * feat: add minimal toolbar mode for config chat input Thread minimalToolbar prop through ChatInputArea → ChatInputContainer → ChatInputBody → ChatInputToolbar → QuickChatContent. When enabled, renders only the submit/cancel button, hiding plan toggle, MCP indicator, context popover, model selector, and other controls. * feat: replace config chat dialog with floating popover panel Replace Dialog with Popover from @kandev/ui so the config chat floats up from the FAB button. Reduce FAB size to h-12/w-12 with larger h-7/w-7 icon. Rename tooltip to "Configuration Chat". Remove standalone ConfigChatFAB component from settings layout. * feat: filter config chat tasks from quick chat listing Add exclude_config query parameter to ListTasksByWorkspace that filters out tasks with metadata.config_mode=true. Apply it in the quick chat sessions API call to prevent config chat tasks from appearing as tabs. * refactor: move config agent profile setting to agents page Remove the Configuration Agent Profile selector from workspace settings and add a dedicated ConfigChatAgentSection to the agents settings page with auto-save on change. * feat: redesign config chat empty state with inline text input Replace clickable template buttons with subtle suggestion hints that fill the textarea on click and disappear once the user starts typing. Add a persistent Textarea + send button at the bottom so users can type directly without clicking templates. This prevents the "no active session" error since QuickChatContent only renders after startSession succeeds. * fix: config chat UX polish — non-clickable suggestions, profile persistence - Make suggestion prompts purely visual (non-interactive text hints) - Update workspace store after saving config agent profile so the profile selector doesn't reappear on subsequent opens - Use useAppStoreApi for store access in callbacks * fix: add inline error display and prevent accidental popover dismiss - Show API errors inline below the input instead of via toast (which can be hidden behind the popover) - Prevent popover from closing on outside click via onInteractOutside, so the user doesn't lose their typed input or active session - Surface error state from useConfigChat hook to the empty state UI * fix: use config-specific placeholder in config chat input Add placeholderOverride prop through ChatInputArea and QuickChatContent so the config chat shows "Ask anything about your configuration..." instead of the default "Continue working on the task..." placeholder. * feat: add default_config_agent_profile_id to workspace DTO and API Wire the field through WorkspaceDTO, FromWorkspace(), HTTP/WS request structs, service layer, and public API types so the config chat default profile persists across page reloads and SSR hydration. * feat: add config chat store actions and prompt delivery via WS Add configChat slice actions (open, close, rename, tab management), pendingPrompt state in useConfigChat hook, and initialPrompt support in QuickChatContent so prompts are delivered via WS after subscription is established rather than via HTTP (avoiding race condition). * fix: resolve dual store issue and message fetch race condition Replace StateProvider with StateHydrator in settings layout to prevent a shadow Zustand store that intercepted WS events. Fix message fetch race condition by merging WS-delivered messages with fetch results instead of replacing them. * feat: make config chat global with command panel integration Add ConfigChatProvider at root layout so the popover is available on all pages. FAB button only visible on settings pages; hidden anchor on other pages for popover positioning. Add "Configuration Chat" command to the command panel (Cmd+K). Add close button and header bar to the popover. * test: add E2E tests for config chat popover and command panel Add tests for opening config chat via FAB on settings page (with full agent response flow) and opening via command panel on kanban page. Fix api-client endpoint and add workspace update fields for E2E setup. * fix: compress props destructuring to satisfy max-lines-per-function lint Rebase merged both sides' additions into ChatInputEditorArea and ChatInputToolbar, pushing them over the 100-line function limit. Use object parameter with compact destructuring to reduce line count. * fix: register config task tools and add exhaustive E2E coverage Config-mode MCP task tools (list_tasks, move_task, delete_task, archive_task) were defined but never registered — registerConfigTaskTools() was never called in registerTools(). Fixed the registration and expanded E2E tests from 5 to 15, covering all testable config-mode MCP tools (workflow, agent, MCP config, and task management) with API-side verification of side effects. * feat: replace create/delete agent with create/delete agent profile in config mode Agent creation/deletion via MCP isn't practical since agents map to registered types. Replace with create_agent_profile and delete_agent_profile tools that manage the per-agent configuration presets users actually need to work with. Wires full chain: agentctl MCP tools → WS actions → backend handlers → controller layer. * chore: fix Prettier formatting * fix: compress props destructuring to satisfy max-lines-per-function lint * test: update MCP tests for agent profile tools and correct config tool count * fix: address CodeRabbit review — normalize MCP mode, set default, consolidate tests * fix: address CodeRabbit review — error leaks, search filter bug, stale prompt, sysprompt tools * feat: expose all workflow step fields in config-mode MCP tools * feat: add workflow CRUD to config-mode MCP tools * fix: split workflow CRUD E2E test into separate tests * fix: prevent cascade deletion of config session in delete workflow E2E test The config-chat endpoint picks the newest workflow (ORDER BY created_at DESC) for the ephemeral task. When the test created a deletable workflow right before starting the config session, the session's own task landed under that workflow. Deleting the workflow then cascaded and wiped the running session's task, causing the MCP tool call to hang. Add a "Config Anchor" bumper workflow so the config session uses it instead of the deletable one. * feat: add executor management to config-mode MCP tools Add 8 new MCP tools for executor and executor profile CRUD in config mode: list/create/update/delete for both executors and profiles. Wired through the full stack (WS actions, backend handlers, agentctl tools, sysprompt) with unit tests and E2E coverage. * fix: auto-focus config chat input and rename popover title * fix: remove executor CRUD MCP tools, keep executor profile management Executors are pre-defined system entities (local, worktree, docker, sprites). Only executor profiles should be managed via config chat. Removes create_executor, update_executor, delete_executor MCP tools while keeping list_executors and all executor profile tools. * fix: add realtime UI sync for workflow steps and agent profiles via MCP Workflow step and agent profile changes made through the config chat MCP tools were not broadcasting events to the frontend. Executor profile events worked because they went through the task service which has event bus integration. - Add WorkflowStep and AgentProfile event types to the event bus - Add WS notification action constants for workflow step events - Publish events from MCP handlers after step/profile CRUD operations - Register gateway subscriptions for the new event types - Add frontend WS handlers for workflow.step.created/updated/deleted * fix: reuse existing config chat session when opened from command panel The command panel always called openConfigChatModal() which reset the session state, causing the profile picker to show again. Now it checks for an existing active session first, matching the FAB behavior. * fix: suppress FAB tooltip while config chat popover is open When closing the popover via the X button, the mouse ends up over the FAB area and the tooltip re-appears. Force the tooltip closed while the popover is open so it cannot re-trigger on close. * fix: thread default_config_agent_profile_id through all workspace data paths The field was defined on the workspace type but never populated in: - SSR mapper (page.tsx) — missing from initial hydration - WS workspace handlers — wiped on workspace.created/updated events - Backend publishWorkspaceEvent — not included in event data This caused the config chat to always show the profile picker when opened from the command panel, since the saved default was lost. * fix: address PR review comments — workspace sync, type fixes, missing tool doc - Sync configChat.workspaceId when promoting next tab in closeConfigChatSession - Add update_task_state_kandev to ConfigContext system prompt - Fix workflow step WS event types (workflow.step.* instead of step.*) - Fix const reassignment in E2E tests (const → let) * refactor: clean up dead code and register missing MCP tools - Register update_task_state, delete_workflow_step, and reorder_workflow_steps as MCP tools in agentctl config mode - Remove dead handleCreateAgent/handleDeleteAgent WS handlers and their ActionMCP constants (only profiles are managed) - Rename openConfigChatModal to startNewConfigChat across frontend - Update sysprompt, tests, and tool counts * fix: remove extra next.config.ts machine * fix: address CI failures — prettier formatting and funlen split * fix: address coderabbit review comments across config-mode stack - Include prompt field in workflow step broadcast payloads - Align sysprompt task-state values with actual backend enum - Return error instead of continuing when MCP mode switch fails - Validate mode values in agentctl set-mcp-mode endpoint - Use errors.Is() for sentinel error comparison - Fix nullish coalescing to allow clearing config agent profile - Expand StepPayload type with missing config-step fields - Derive config agent Select value from store instead of local state * fix: reuse persisted config chat session and reduce toolbar complexity - Look up configChat.sessions for workspace before creating new session - Extract MinimalToolbar and move defaults to module-level object to reduce ChatInputToolbar cyclomatic complexity from 16 to 7 * fix: E2E executor test strict mode violation and invalid cleanup - Use exact match for list_executors text to avoid matching 2 elements - Remove cleanup that tried to delete pre-existing system executor - Reuse persisted config chat session for workspace in global commands - Reduce ChatInputToolbar cyclomatic complexity with module-level defaults	2026-03-13 00:44:06 +00:00
Carlos Florêncio	37e6b1bd6d	fix: recover from agent MCP timeout during clarification wait (#413)	2026-03-10 22:48:30 +00:00
Carlos Florêncio	c73fa7471b	fix: resolve clarification MCP timeout with cancel-and-resume flow (#362)	2026-03-05 00:18:57 +00:00
Carlos Florêncio	2ae26add46	feat: improve vscode cleanup (#294)	2026-02-26 13:34:47 +00:00
Carlos Florêncio	61ef224ec7	style: fix format issues (#255)	2026-02-19 22:18:01 +00:00
Carlos Florêncio	fd4ae1eb54	fix: resolve all backend golangci-lint violations (#245)	2026-02-18 18:24:24 +00:00
Carlos Florêncio	122deb3b13	feat: add GitHub Copilot CLI and Sourcegraph Amp agent support (#130)	2026-02-01 00:40:56 +00:00
João Salavisa	6eb24006f9	feat: implement task plans feature (#131)	2026-01-31 00:51:17 +00:00
Carlos Florêncio	3f7154e186	feat: chat messages normalized (#121)	2026-01-30 17:16:05 +00:00
João Salavisa	f229c72393	feat: migrate MCP server from backend to agentctl (#124)	2026-01-30 15:15:19 +00:00
João Salavisa	49f98b88ff	fix: Multiple bug fixes for agent lifecycle and task creation (#122)	2026-01-29 23:40:02 +00:00
Carlos Florêncio	752e6b4655	feat: process runners (#71)	2026-01-22 16:39:13 +00:00
João Salavisa	c57a82f81d	feat(mcp): add embedded MCP server with dual transport support (#67)	2026-01-21 12:03:46 +00:00
João Salavisa	7e43534cd2	refactor: remove auggie-specific code and use standard ACP protocol (#47) ... - Remove ProtocolAuggie constant and AuggieAdapter - Remove hooks system (HooksRegistry, AgentHooks, AuggieHooks) - Remove PreGeneratedSessionID and SessionDir fields - Change auggie-agent to use standard 'acp' protocol in agents.json - Add model select dropdown to profile settings page - Fix unused variable lint warnings Auggie now uses the standard ACP protocol flow like other agents. The --resume flag was found to not work in ACP mode, so the workaround was unnecessary since session/new returns valid session IDs.	2026-01-19 22:41:00 +00:00
João Salavisa	18f4b4e6fb	Cleanup deprecated code and fix session recovery (#37)	2026-01-18 15:58:56 +00:00
João Salavisa	2e108b040f	refactor: session resumption cleanup and AgentInstance → AgentExecution rename (#23) ... * feat: session resumption with ACP session ID support - Add useSessionResumption hook for automatic session recovery on page reload - Fix ACPSessionID not being set on AgentInstance in Launch() and createInstance() - Add task.session.status and task.session.resume WebSocket actions - Fix shell terminal retry logic for resumed sessions - Fix taskSessionState reading from Zustand store instead of local state - Remove dead code: unused stream handling functions in lifecycle/manager.go - Fix test mock interfaces (CleanupStaleInstanceByTaskID, NewManager signature) - Add enhanced logging for session resumption debugging - Simplify page-client.tsx state merging logic * refactor: remove unused AgentController (dead code) - Delete agent_controller.go - entirely unused duplicate of Manager methods - Remove agentController field from Manager struct - Remove AgentController initialization in NewManager The AgentController duplicated StartAgentProcess, initializeACPSession, and PromptAgent from Manager but was never called. All agent control operations go through Manager's own methods. * refactor: remove unused MemoryRepository (dead code) - Delete memory.go - entire MemoryRepository implementation unused (replaced by SQLite repository) - Delete memory_test.go - tests for removed MemoryRepository Found via 'deadcode' tool analysis. * refactor: consolidate agentctl packages into server/client structure Reorganize the confusing dual agentctl packages: - internal/agentctl/* → internal/agentctl/server/ - internal/agent/agentctl/* → internal/agentctl/client/ Structure is now: internal/agentctl/ ├── client/ - HTTP client for talking to agentctl servers │ ├── client.go │ ├── standalone.go │ ├── acp.go │ └── launcher/ ├── server/ - agentctl server implementation (runs in containers) │ ├── api/ │ ├── adapter/ │ ├── acp/ │ ├── config/ │ ├── instance/ │ ├── process/ │ └── shell/ └── types/ - Shared types used by both client and server Also consolidated shared types (GitStatusUpdate, FileInfo, etc.) from process/types.go into the shared types/ package. * refactor: rename AgentInstance to AgentExecution for clarity Rename AgentInstance -> AgentExecution across the entire codebase to better reflect what it represents: an agent actively working on a task, not just a generic 'instance'. Changes: - AgentInstance struct -> AgentExecution - InstanceStore -> ExecutionStore (new file: execution_store.go) - RecoveredInstance -> RecoveredExecution - All method names: GetInstance* -> GetExecution*, ListInstances -> ListExecutions, etc. - All parameters: instanceID -> executionID - All local variables: instance -> execution - DTOs and JSON tags: agent_instance_id -> agent_execution_id - Database columns: agent_instance_id -> agent_execution_id Also consolidates duplicate prompt logic: - Merge sendInitialPrompt and SendPrompt into single SendPrompt method - Add validateStatus parameter to differentiate initial vs follow-up prompts Also extracts ACP session management to session.go: - Move initializeACPSession logic to SessionManager.InitializeAndPrompt - Move PromptAgent logic to SessionManager.SendPrompt - Manager now delegates to SessionManager for all prompt operations Also removes duplicate event publishing: - Remove 9 duplicate m.publish*() methods from manager.go - Use existing EventPublisher methods via m.eventPublisher.Publish*() Also extracts types to dedicated file: - Move AgentExecution, LaunchRequest, PromptResult, etc. to types.go Net reduction: ~600 lines removed from manager.go through deduplication and better separation of concerns. * fix(backend): Docker runtime now creates instance via ControlServer The Docker runtime was incorrectly creating an agentctl.Client pointing directly to port 9999 (the ControlServer port), then trying to call instance-specific APIs like /api/v1/agent/configure which only exist on per-instance servers. Fixed by making Docker runtime follow the same flow as standalone mode: - LaunchContainer now uses StandaloneCtl to create an instance via the control API, then creates Client pointing to the instance port - RecoverInstances queries the ControlServer to get the instance port - Added waitForHealth helper to wait for agentctl to be ready Also improved HTTP error handling in agentctl client: - All client methods now check HTTP status codes before JSON decoding - Better error messages include status code and response body * fix: restore session resumption imports and variables * fix: add sessionId prop back to TaskLayout	2026-01-16 22:14:19 +00:00