cowork-os 0.3.71 → 0.3.73

package/CHANGELOG.md

@@ -7,6 +7,34 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 
 ## [Unreleased]
 
+## [0.3.73] - 2026-02-14
+
+### Fixed
+- **Release pipeline fix**: included daemon TypeScript sources in shared ESLint targets so `npm run lint` runs instead of failing with parse errors before build/publish steps.
+- **Workspace/task validation fix**: enforced `PersonalityId` validation for task agent configs to prevent runtime/inference mismatches during task creation.
+- **CLI and release install tests alignment**: updated control-plane and skill validation tests to match current runtime behaviors and skill metadata output.
+- **Workspace preflight reliability**: stabilized ambiguous temp-task auto-switch behavior when project-signals are present and tests now validate that behavior.
+
+## [0.3.72] - 2026-02-14
+
+### Added
+- **Session workspace isolation and cleanup**: temp tasks now get session-scoped workspace IDs, dedicated temp directories, and automatic pruning by age + usage caps.
+- **Autonomous task mode** in execution flow and control-plane/web-UI paths, with optional bypass of interactive approval prompts where explicitly enabled.
+- **Companion-mode handling for short conversational prompts** to return concise check-in responses without running task pipeline when appropriate.
+- **Search execution ordering** now prefers Brave when available and can safely fallback through configured providers automatically.
+- **PDF parsing compatibility wrapper** with runtime-safe handling for both legacy and v2 parser module shapes.
+
+### Changed
+- **Task completion validation tightened** with final-response contracts (required direct answers, artifact checks, verification evidence).
+- **Stricter tool failure handling** for hard/unavailable/disallowed outcomes to prevent false completion without real progress.
+- **Temporary workspace handling** now uses explicit session-aware IDs and filters temp workspaces from user-visible lists consistently.
+- **Search and file tools** now enforce more bounded scanning behavior and clearer fallback behavior under high-load conditions.
+
+### Fixed
+- **Watch/skip recommendation tasks** now block artifact tools and require direct recommendation output.
+- **Intermittent approval/partial-task updates** reduced by normalizing auto-approved events in UI and task-stream handling.
+- **Temp workspace lifecycle reliability** improved through scheduled pruning and safer restore/create paths.
+
 ## [0.3.69] - 2026-02-11
 
 ### Fixed
@@ -481,6 +509,8 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 
 | Version | Date | Highlights |
 |---------|------|------------|
+| 0.3.73 | 2026-02-14 | Release automation hardening and task/workspace validation fixes |
+| 0.3.72 | 2026-02-14 | Session-based temp workspaces, autonomous execution mode, safer completion validation |
 | 0.3.29 | 2025-02-08 | Multi-provider image generation, visual annotation, local embeddings, verification UX |
 | 0.3.25 | 2025-02-05 | Google Workspace integration, gateway enhancements, agent retry logic |
 | 0.1.6 | 2025-01-25 | Discord bot integration with slash commands |
@@ -490,7 +520,10 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 | 0.1.0 | 2025-01-24 | First public release with core features |
 | 0.0.1 | 2025-01-20 | Initial development setup |
 
-[Unreleased]: https://github.com/CoWork-OS/CoWork-OS/compare/v0.3.29...HEAD
+[Unreleased]: https://github.com/CoWork-OS/CoWork-OS/compare/v0.3.73...HEAD
+[0.3.73]: https://github.com/CoWork-OS/CoWork-OS/releases/tag/v0.3.73
+[0.3.72]: https://github.com/CoWork-OS/CoWork-OS/releases/tag/v0.3.72
+[0.3.71]: https://github.com/CoWork-OS/CoWork-OS/releases/tag/v0.3.71
 [0.3.29]: https://github.com/CoWork-OS/CoWork-OS/releases/tag/v0.3.29
 [0.3.25]: https://github.com/CoWork-OS/CoWork-OS/releases/tag/v0.3.25
 [0.1.6]: https://github.com/CoWork-OS/CoWork-OS/releases/tag/v0.1.6

package/README.md

@@ -36,6 +36,13 @@ Your AI needs a secure home. CoWork OS provides the runtime, security layers, an
 | **Security-First** | 2800+ unit tests, configurable guardrails, approval workflows, gateway hardening |
 | **Local-First** | Your data stays on your machine. BYOK (Bring Your Own Key) |
 
+### What’s new in 0.3.73
+
+- **Release and install reliability**: fixed publish-blocking type/lint regressions so npm desktop/client releases can be published reliably.
+- **Task config validation**: stricter `personalityId` validation for task `agentConfig` avoids malformed IDs reaching task execution.
+- **Control-plane/LLM error handling alignment**: tests and runtime error codes now match shared validation shapes for predictable client behavior.
+- **Workspace auto-switch confidence**: improved ambiguous temp-task preflight handling when preferred workspace signals are available.
+
 > **Status**: macOS desktop app + headless/server mode (Linux/VPS). Cross-platform desktop support planned.
 
 ---
@@ -70,6 +77,16 @@ pkill -f '/cowork-os' || true
 npx cowork-os
 ```
 
+#### Install reliability notes (macOS / low-memory environments)
+
+- If install fails with `SIGKILL` during `node_modules/electron/install.js`, use a two-step install:
+  - `npm install --ignore-scripts cowork-os@latest --no-audit --no-fund`
+  - `npm run setup` (from the install directory) before launching the CLI
+- For local package testing, use the same `--ignore-scripts` flow with the tarball:
+  - `npm init -y`  
+  - `npm install --ignore-scripts /path/to/cowork-os-<version>.tgz`
+- If you already have a global install, verify with `coworkd-node --version` and avoid launching without dependency setup on first run.
+
 You can also install globally and launch directly:
 
 ```bash
@@ -193,6 +210,7 @@ ssh -N -L 28789:127.0.0.1:18789 user@your-vps
 
 - `http://127.0.0.1:18789/` (or `http://127.0.0.1:28789/` if you used 28789)
 - Paste the token from step 1
+- If you skipped `OPENAI_API_KEY`/`ANTHROPIC_API_KEY`, use the **LLM Setup** panel in this UI to configure provider credentials.
 
 For production/persistent setup and Docker/systemd options, use:
 
@@ -407,7 +425,6 @@ Configure in **Settings** > **Appearance**.
 ### Agent Capabilities
 
 - **Task-Based Workflow**: Multi-step execution with plan-execute-observe loops
-- **Goal Mode**: Define success criteria and auto-retry until verification passes
 - **Dynamic Re-Planning**: Agent can revise its plan mid-execution
 - **85+ Built-in Skills**: GitHub, Slack, Notion, Spotify, Apple Notes, and more
 - **Document Creation**: Excel, Word, PDF, PowerPoint with professional formatting
@@ -417,10 +434,13 @@ Configure in **Settings** > **Appearance**.
 - **Performance Reviews**: Score and review agent-role outcomes, with autonomy-level recommendations
 - **Voice Calls**: Outbound phone calls via ElevenLabs Agents (list agents, list numbers, initiate calls)
 - **Vision**: Analyze workspace images (screenshots, photos, diagrams) via `analyze_image` tool (OpenAI, Anthropic, or Gemini)
+- **X Browser Fallback**: `x_action` automatically falls back to browser-mode read/write flows when Bird CLI is blocked (rate limits, auth challenges, access issues)
+- **Attachment OCR (Optional)**: Extract image text in local file previews with Tesseract (`tesseract` binary must be installed and on `PATH`)
 - **Image Generation**: Create images via `generate_image` with multi-provider support (Gemini, OpenAI gpt-image-1/1.5/DALL-E, Azure OpenAI) and automatic provider selection
 - **Visual Annotation**: Iterative image refinement with the Visual Annotator — generate, annotate, refine, repeat until approved
 - **Email IMAP Access**: Direct IMAP mailbox access via `email_imap_unread` — check unread emails without needing Google Workspace
 - **Workspace Recency**: Workspaces ordered by last used time for quick access
+- **Hook-Triggered Heartbeat Wakeups**: Agents can act on hook-based wake events as explicit check-in prompts
 
 ### Voice Mode
 
@@ -489,6 +509,40 @@ Notes:
 
 Configure in **Settings** > **Memory Hub**.
 
+### Role Profile Files (`.cowork/agents/`)
+
+You can define per-role personality and operating guidelines directly in workspace files.  
+When a role runs, CoWork loads these first; if missing, it falls back to legacy role metadata.
+
+Expected structure:
+
+| File | Purpose |
+|---|---|
+| `.cowork/agents/<role-id>/SOUL.md` | Role personality, behavior style, and execution philosophy |
+| `.cowork/agents/<role-id>/IDENTITY.md` | Role-specific identity and constraints |
+| `.cowork/agents/<role-id>/RULES.md` | Operational rules, safety boundaries, communication defaults |
+
+Role resolution checks these folder candidates in order:
+
+- `<normalized role.name>` (keeps folder-safe letters, numbers, `_`, `-`, `.`)
+- slugified variant of role name (ASCII-safe fallback)
+- `<normalized role.displayName>` (same normalization)
+- slugified variant of displayName
+- `<normalized role.id>`
+- slugified variant of id
+- `default`
+
+Only non-empty `.cowork/agents/...` files are used. If no files are found, CoWork uses existing DB role `soul` notes.
+
+Tips:
+
+- Start by cloning a profile from an existing role: copy one `*.md` file to a new role folder.
+- Keep templates simple; the first 4k chars per file are injected into prompts.
+- Keep sensitive values out of these files; `.cowork` content is sanitized before use.
+- Quick scaffold:
+  - `mkdir -p .cowork/agents/<role-id> && printf '%s\n' "# SOUL.md" "..." > .cowork/agents/<role-id>/SOUL.md`
+  - If you are unsure about the folder name, place a `default` profile at `.cowork/agents/default/` and copy it to `.../<role-id>/`.
+
 ### Agent Teams
 
 Coordinate multiple agents working together on complex tasks with shared state.
@@ -547,6 +601,10 @@ Full Playwright integration:
 - Click, fill forms, type text, press keys
 - Extract page content, links, and form data
 - Scroll pages, wait for elements, execute JavaScript
+- Browser automation supports `browser_channel`:
+  - `chromium` (default bundled Chromium)
+  - `chrome` (system Google Chrome)
+  - `brave` (system Brave or a path set in `BRAVE_PATH`)
 
 ### System Tools
 
@@ -1680,7 +1738,7 @@ Claude Code-style tools for developers.
 → grep pattern="TODO:" glob="*.ts"
 ```
 
-**Smart Document Detection**: Automatically detects document-heavy workspaces (PDF/DOCX) and provides helpful guidance to use `read_file` instead, since grep only searches text files.
+**Smart Document Detection**: Automatically detects document-heavy workspaces (PDF/DOCX/PPTX) and provides helpful guidance to use `read_file` instead, since grep only searches text files.
 
 ### edit_file - Surgical Editing
 
@@ -2207,6 +2265,7 @@ Users must comply with their model provider's terms:
 - [x] Chat commands: `/schedule`, `/digest`, `/followups`, `/brief` across all gateway channels
 - [x] Inbound attachment persistence (channel messages save files to `.cowork/inbox/attachments/`)
 - [x] Cron template variables (`{{today}}`, `{{chat_messages}}`, etc.) and conditional delivery
+- [x] Brave browser channel support via `browser_navigate` (`browser_channel: "chrome" | "chromium" | "brave"`)
 - [x] Image generation via `generate_image` with multi-provider support (Gemini, OpenAI, Azure OpenAI)
 - [x] Visual annotation tools and Agentic Image Loop skill for iterative image refinement
 - [x] Inline image preview in task event timeline
@@ -2219,6 +2278,7 @@ Users must comply with their model provider's terms:
 - [x] Control plane web dashboard, approval API, and channel management API
 - [x] Docker and VPS deployment support (Dockerfiles, docker-compose, systemd units)
 - [x] Dedicated workspaces for scheduled cron jobs (auto-created from temp workspace)
+- [x] Hook wake requests for heartbeat processing, including queueing, coalescing, and immediate scheduling
 
 ### Planned
 

package/dist/daemon/daemon/control-plane-methods.js

@@ -46,8 +46,8 @@ const protocol_1 = require("../electron/control-plane/protocol");
 const settings_1 = require("../electron/control-plane/settings");
 const types_1 = require("../shared/types");
 const repositories_1 = require("../electron/database/repositories");
-const llm_1 = require("../electron/agent/llm");
 const search_1 = require("../electron/agent/search");
+const llm_configure_1 = require("../electron/control-plane/llm-configure");
 const runtime_mode_1 = require("../electron/utils/runtime-mode");
 const user_data_dir_1 = require("../electron/utils/user-data-dir");
 function requireScope(client, scope) {
@@ -844,6 +844,11 @@ function registerControlPlaneMethods(server, deps) {
         await channelGateway.removeChannel(channelId);
         return { ok: true };
     });
+    // LLM setup (headless-friendly credential/provider configuration).
+    server.registerMethod(protocol_1.Methods.LLM_CONFIGURE, async (client, params) => {
+        requireScope(client, 'admin');
+        return (0, llm_configure_1.configureLlmFromControlPlaneParams)(params);
+    });
     // Config/health (sanitized; no secrets).
     server.registerMethod(protocol_1.Methods.CONFIG_GET, async (client) => {
         requireScope(client, 'read');
@@ -863,12 +868,7 @@ function registerControlPlaneMethods(server, deps) {
                 tasksByStatus[status] = safeCount;
             taskTotal += safeCount;
         }
-        const llmStatus = llm_1.LLMProviderFactory.getConfigStatus();
-        const llm = {
-            currentProvider: llmStatus.currentProvider,
-            currentModel: llmStatus.currentModel,
-            providers: llmStatus.providers,
-        };
+        const llm = (0, llm_configure_1.getControlPlaneLlmStatus)();
         const anyLlmConfigured = llm.providers.some((p) => p.configured);
         const currentProviderConfigured = llm.providers.find((p) => p.type === llm.currentProvider)?.configured || false;
         const searchStatus = search_1.SearchProviderFactory.getConfigStatus();
@@ -896,7 +896,7 @@ function registerControlPlaneMethods(server, deps) {
             warnings.push('No workspaces configured. Set COWORK_BOOTSTRAP_WORKSPACE_PATH on startup or create one via workspace.create.');
         }
         if (!anyLlmConfigured) {
-            warnings.push('No LLM provider credentials configured. Set COWORK_IMPORT_ENV_SETTINGS=1 (or run with --import-env-settings) plus an API key (e.g. OPENAI_API_KEY), then restart.');
+            warnings.push('No LLM provider credentials configured. Configure one via Control Plane (LLM Setup / llm.configure), or use COWORK_IMPORT_ENV_SETTINGS=1 with provider env vars and restart.');
         }
         else if (!currentProviderConfigured) {
             warnings.push(`Selected LLM provider "${llm.currentProvider}" is not configured. Either switch provider or configure its credentials.`);

package/dist/daemon/electron/agent/browser/browser-service.js

@@ -50,9 +50,42 @@ class BrowserService {
         this.options = {
             headless: options.headless ?? true,
             timeout: options.timeout ?? 30000,
-            viewport: options.viewport ?? { width: 1280, height: 720 }
+            viewport: options.viewport ?? { width: 1280, height: 720 },
+            userDataDir: options.userDataDir,
+            channel: options.channel,
         };
     }
+    async resolveBraveExecutablePath() {
+        const envPath = process.env.BRAVE_PATH?.trim();
+        const candidates = [
+            envPath,
+            process.platform === 'darwin'
+                ? '/Applications/Brave Browser.app/Contents/MacOS/Brave Browser'
+                : undefined,
+            process.platform === 'linux' ? '/usr/bin/brave-browser' : undefined,
+            process.platform === 'linux' ? '/usr/bin/brave-browser-stable' : undefined,
+            process.platform === 'linux' ? '/snap/bin/brave' : undefined,
+            process.platform === 'win32' && process.env.LOCALAPPDATA
+                ? path.join(process.env.LOCALAPPDATA, 'BraveSoftware', 'Brave-Browser', 'Application', 'brave.exe')
+                : undefined,
+            process.platform === 'win32'
+                ? 'C:\\Program Files\\BraveSoftware\\Brave-Browser\\Application\\brave.exe'
+                : undefined,
+            process.platform === 'win32'
+                ? 'C:\\Program Files (x86)\\BraveSoftware\\Brave-Browser\\Application\\brave.exe'
+                : undefined,
+        ].filter((value) => Boolean(value));
+        for (const candidate of candidates) {
+            try {
+                await fs.access(candidate);
+                return candidate;
+            }
+            catch {
+                // Keep scanning candidates.
+            }
+        }
+        return undefined;
+    }
     /**
      * Initialize the browser
      * Uses try-finally to ensure cleanup on errors
@@ -64,11 +97,19 @@ class BrowserService {
         let context = null;
         try {
             const channel = this.options.channel === 'chrome' ? 'chrome' : undefined;
+            const executablePath = this.options.channel === 'brave'
+                ? await this.resolveBraveExecutablePath()
+                : undefined;
+            if (this.options.channel === 'brave' && !executablePath) {
+                throw new Error('Brave browser was requested but no Brave executable was found. ' +
+                    'Install Brave or set BRAVE_PATH to the Brave binary path.');
+            }
             if (this.options.userDataDir) {
                 await fs.mkdir(this.options.userDataDir, { recursive: true });
                 context = await playwright_1.chromium.launchPersistentContext(this.options.userDataDir, {
                     headless: this.options.headless,
                     ...(channel ? { channel } : {}),
+                    ...(executablePath ? { executablePath } : {}),
                     viewport: this.options.viewport,
                 });
                 browser = context.browser();
@@ -77,6 +118,7 @@ class BrowserService {
                 browser = await playwright_1.chromium.launch({
                     headless: this.options.headless,
                     ...(channel ? { channel } : {}),
+                    ...(executablePath ? { executablePath } : {}),
                 });
                 context = await browser.newContext({
                     viewport: this.options.viewport,

package/dist/daemon/electron/agent/custom-skill-loader.js

@@ -304,7 +304,8 @@ class CustomSkillLoader {
      * List skills that can be automatically invoked by the model
      * Excludes guidelines and skills with disableModelInvocation set
      */
-    listModelInvocableSkills() {
+    listModelInvocableSkills(options = {}) {
+        const availableToolNames = options.availableToolNames;
         return this.listSkills().filter((skill) => {
             // Exclude guideline skills
             if (skill.type === 'guideline')
@@ -315,6 +316,22 @@ class CustomSkillLoader {
             // Exclude skills that explicitly disable model invocation
             if (skill.invocation?.disableModelInvocation === true)
                 return false;
+            // If tool availability is provided, filter out skills that cannot run in this context.
+            if (availableToolNames) {
+                const skillRequires = (skill.requires || {});
+                const requiredTools = Array.isArray(skillRequires.tools)
+                    ? skillRequires.tools.filter((tool) => typeof tool === 'string' && tool.trim().length > 0)
+                    : [];
+                if (requiredTools.some((tool) => !availableToolNames.has(tool))) {
+                    return false;
+                }
+                // Skills requiring external binaries generally need run_command access.
+                const hasBinaryRequirements = (Array.isArray(skill.requires?.bins) && skill.requires.bins.length > 0) ||
+                    (Array.isArray(skill.requires?.anyBins) && skill.requires.anyBins.length > 0);
+                if (hasBinaryRequirements && !availableToolNames.has('run_command')) {
+                    return false;
+                }
+            }
             return true;
         });
     }
@@ -322,8 +339,8 @@ class CustomSkillLoader {
      * Get formatted skill descriptions for the model's system prompt
      * Groups skills by category and includes parameter info
      */
-    getSkillDescriptionsForModel() {
-        const skills = this.listModelInvocableSkills();
+    getSkillDescriptionsForModel(options = {}) {
+        const skills = this.listModelInvocableSkills(options);
         if (skills.length === 0) {
             return '';
         }
@@ -345,10 +362,47 @@ class CustomSkillLoader {
                     ? ` (params: ${skill.parameters.map(p => p.name + (p.required ? '*' : '')).join(', ')})`
                     : '';
                 lines.push(`- ${skill.id}: ${skill.description}${paramInfo}`);
+                const routingHints = this.getSkillRoutingHints(skill);
+                for (const hint of routingHints) {
+                    lines.push(`  ${hint}`);
+                }
             }
         }
         return lines.join('\n');
     }
+    /**
+     * Build compact routing and success hints for model prompt listing.
+     * These are intentionally short and should act like decision boundaries.
+     */
+    getSkillRoutingHints(skill) {
+        const routing = skill.metadata?.routing;
+        if (!routing) {
+            return [];
+        }
+        const hints = [];
+        if (routing.useWhen) {
+            hints.push(`Use when: ${routing.useWhen}`);
+        }
+        if (routing.dontUseWhen && !this.isLowSignalRoutingHint('dontUseWhen', routing.dontUseWhen)) {
+            hints.push(`Don't use when: ${routing.dontUseWhen}`);
+        }
+        if (routing.outputs && !this.isLowSignalRoutingHint('outputs', routing.outputs)) {
+            hints.push(`Outputs: ${routing.outputs}`);
+        }
+        if (routing.successCriteria && !this.isLowSignalRoutingHint('successCriteria', routing.successCriteria)) {
+            hints.push(`Success criteria: ${routing.successCriteria}`);
+        }
+        if (routing.expectedArtifacts?.length) {
+            hints.push(`Artifacts: ${routing.expectedArtifacts.join(', ')}`);
+        }
+        return hints;
+    }
+    isLowSignalRoutingHint(kind, value) {
+        const normalized = value.trim();
+        if (!normalized)
+            return true;
+        return CustomSkillLoader.LOW_SIGNAL_ROUTING_HINT_PATTERNS[kind].some((pattern) => pattern.test(normalized));
+    }
     /**
      * Get a specific skill by ID
      */
@@ -358,8 +412,8 @@ class CustomSkillLoader {
     /**
      * Expand a skill's prompt template with parameter values
      */
-    expandPrompt(skill, parameterValues) {
-        let prompt = this.expandBaseDir(skill.prompt, skill);
+    expandPrompt(skill, parameterValues, context = {}) {
+        let prompt = this.expandSkillPromptPlaceholders(skill.prompt, skill, context);
         // Replace {{param}} placeholders with values
         if (skill.parameters) {
             for (const param of skill.parameters) {
@@ -382,6 +436,13 @@ class CustomSkillLoader {
         const baseDir = this.resolveBaseDir(skill);
         return prompt.replace(/\{baseDir\}/g, baseDir);
     }
+    expandSkillPromptPlaceholders(prompt, skill, context) {
+        let output = this.expandBaseDir(prompt, skill);
+        if (context.artifactDir) {
+            output = output.replace(/\{artifactDir\}/g, context.artifactDir);
+        }
+        return output;
+    }
     resolveBaseDir(skill) {
         const fileDir = skill.filePath ? path.dirname(skill.filePath) : this.bundledSkillsDir;
         const candidates = [
@@ -580,6 +641,18 @@ class CustomSkillLoader {
     }
 }
 exports.CustomSkillLoader = CustomSkillLoader;
+CustomSkillLoader.LOW_SIGNAL_ROUTING_HINT_PATTERNS = {
+    dontUseWhen: [
+        /planning documents,\s*high-level strategy,\s*or non-executable discussion/i,
+    ],
+    outputs: [
+        /^Outcome from .*task-specific result plus concrete action notes\.?$/i,
+    ],
+    successCriteria: [
+        /returns concrete actions and decisions matching the requested task/i,
+        /no fabricated tool-side behavior/i,
+    ],
+};
 // Singleton instance
 let instance = null;
 function getCustomSkillLoader(config) {

package/dist/daemon/electron/agent/daemon.js

@@ -381,6 +381,11 @@ class AgentDaemon extends events_1.EventEmitter {
         const parent = this.taskRepo.findById(params.parentTaskId);
         const parentGatewayContext = parent?.agentConfig?.gatewayContext;
         const childGatewayContext = params.agentConfig?.gatewayContext;
+        const parentAutonomousMode = parent?.agentConfig?.autonomousMode === true;
+        const mergedAutonomousMode = parentAutonomousMode || params.agentConfig?.autonomousMode === true;
+        const mergedAllowUserInput = mergedAutonomousMode
+            ? false
+            : params.agentConfig?.allowUserInput ?? parent?.agentConfig?.allowUserInput;
         // Prevent privilege escalation: a child task may not become "more private" than its parent.
         const mergedGatewayContext = (() => {
             const rank = {
@@ -418,6 +423,12 @@ class AgentDaemon extends events_1.EventEmitter {
             if (mergedToolRestrictions) {
                 next.toolRestrictions = mergedToolRestrictions;
             }
+            if (mergedAutonomousMode !== undefined) {
+                next.autonomousMode = mergedAutonomousMode;
+            }
+            if (mergedAllowUserInput !== undefined) {
+                next.allowUserInput = mergedAllowUserInput;
+            }
             return Object.keys(next).length > 0 ? next : undefined;
         })();
         const task = this.taskRepo.create({
@@ -520,7 +531,15 @@ class AgentDaemon extends events_1.EventEmitter {
             return;
         const planSummary = this.buildPlanSummary(plan);
         for (const role of rolesToDispatch) {
-            const childPrompt = (0, agent_dispatch_1.buildAgentDispatchPrompt)(role, { title: task.title, prompt: task.prompt }, { ...(planSummary ? { planSummary } : {}), includeRoleDetails: false });
+            const workspacePath = task.workspaceId
+                ? this.workspaceRepo.findById(task.workspaceId)?.path
+                : undefined;
+            const childPrompt = (0, agent_dispatch_1.buildAgentDispatchPrompt)(role, { title: task.title, prompt: task.prompt }, {
+                ...(planSummary ? { planSummary } : {}),
+                includeRoleDetails: false,
+                includeRoleProfile: true,
+                workspacePath,
+            });
             const childTask = await this.createChildTask({
                 title: `@${role.displayName}: ${task.title}`,
                 prompt: childPrompt,
@@ -714,6 +733,27 @@ class AgentDaemon extends events_1.EventEmitter {
      * Request approval from user for an action
      */
     async requestApproval(taskId, type, description, details) {
+        const task = this.taskRepo.findById(taskId);
+        if (task?.agentConfig?.autonomousMode) {
+            const approval = this.approvalRepo.create({
+                taskId,
+                type: type,
+                description,
+                details,
+                status: 'approved',
+                requestedAt: Date.now(),
+            });
+            this.approvalRepo.update(approval.id, 'approved');
+            this.logEvent(taskId, 'approval_requested', {
+                approval,
+                autoApproved: true,
+            });
+            this.logEvent(taskId, 'approval_granted', {
+                approvalId: approval.id,
+                autoApproved: true,
+            });
+            return true;
+        }
         const approval = this.approvalRepo.create({
             taskId,
             type: type,
@@ -1643,6 +1683,16 @@ class AgentDaemon extends events_1.EventEmitter {
     getWorkspaceByPath(path) {
         return this.workspaceRepo.findByPath(path);
     }
+    /**
+     * Get the most recently used non-temporary workspace, if any.
+     */
+    getMostRecentNonTempWorkspace() {
+        const workspaces = this.workspaceRepo.findAll();
+        return workspaces.find((workspace) => !(0, types_1.isTempWorkspaceId)(workspace.id) &&
+            !workspace.isTemp &&
+            typeof workspace.path === 'string' &&
+            workspace.path.trim().length > 0);
+    }
     /**
      * Create a new workspace with default permissions
      */
@@ -1657,7 +1707,7 @@ class AgentDaemon extends events_1.EventEmitter {
         return this.workspaceRepo.create(name, path, defaultPermissions);
     }
     /**
-     * Update task fields (for Goal Mode attempt tracking, etc.)
+     * Update task fields (for retry/verification attempt tracking, etc.)
      */
     updateTask(taskId, updates) {
         const existing = this.taskRepo.findById(taskId);