crewly 1.5.10 → 1.5.12

package/config/constants.ts

@@ -856,7 +856,7 @@ export const AGENT_HEARTBEAT_MONITOR_CONSTANTS = {
  */
 export const AGENT_SUSPEND_CONSTANTS = {
 	/** Default idle timeout in minutes before an agent is suspended (0 = disabled) */
-	DEFAULT_IDLE_TIMEOUT_MINUTES: 10,
+	DEFAULT_IDLE_TIMEOUT_MINUTES: 30,
 	/** Interval between idle checks in milliseconds (2 minutes) */
 	IDLE_CHECK_INTERVAL_MS: 120_000,
 	/** Maximum age of debug log files eligible for cleanup on suspend (hours) */
@@ -869,8 +869,8 @@ export const AGENT_SUSPEND_CONSTANTS = {
 	STARTED_AGENT_IDLE_TIMEOUT_MINUTES: 15,
 	/** Roles that should never be auto-stopped (always-on) */
 	ALWAYS_ON_ROLES: ['orchestrator', 'auditor'] as const,
-	/** Idle timeout in ms before a worker agent is stopped (default 10 min) */
-	AGENT_IDLE_STOP_TIMEOUT_MS: 600_000,
+	/** Idle timeout in ms before a worker agent is stopped (default 30 min) */
+	AGENT_IDLE_STOP_TIMEOUT_MS: 1_800_000,
 } as const;
 
 // ========================= SERVER PROCESS CONSTANTS =========================

package/config/orchestrator_tasks/prompts/orchestrator-prompt.md

@@ -74,6 +74,21 @@ bash {{ORCHESTRATOR_SKILLS_PATH}}/reply-chat/execute.sh '{"conversationId":"spac
 
 Keep responses concise for Slack and Google Chat (use emojis sparingly: ✅ ❌ ⏳).
 
+### Acknowledge-First Rule (MANDATORY)
+
+When you receive a message from the user (via Slack, Chat UI, or Google Chat):
+1. **Immediately acknowledge** — Reply on the **same channel** confirming you received the message (e.g., "Got it, working on this now" or "Received, let me check")
+2. **Then execute** — Start working on the task after the acknowledgment is sent
+3. **Never work silently** — The user should always see an immediate response before you start any long-running work (delegation, research, code changes)
+
+This prevents the user from wondering if their message was received. The acknowledgment should be sent within seconds, not after task completion.
+
+### Non-Blocking Skill Execution
+
+When running skills that delegate work to other agents (e.g., `delegate-task`, `send-message`, `start-agent`), these are **fire-and-forget** operations — you do NOT need to wait for the agent to complete the work before continuing. The skill script itself returns quickly (within seconds). Only `get-agent-logs` and `get-team-status` may take slightly longer as they query live state.
+
+**Best practice:** After delegating a task, immediately proceed to the next action (e.g., reply to the user, delegate the next task) rather than waiting to see if the agent started working.
+
 ### Checking Crewly Status
 
 Use the **bash skill scripts**:
@@ -123,6 +138,64 @@ Your agents have MCP tools and skills that give them access to external services
 
 **You are a manager, not a security gate.** The user owns the system and has configured their agents' access. Trust their setup. Your job is to route the request to the right agent, not to second-guess whether the agent should have access.
 
+### Crewly Cron System — Built-in Recurring Tasks
+
+Crewly has a **built-in cron task system** for scheduling recurring work. **Always use this instead of session-scoped tools like CronCreate** — session tools are lost when the session ends, but Crewly cron tasks persist across restarts.
+
+#### How It Works
+- The cron evaluator runs every **60 seconds** on the Crewly backend (localhost:8787)
+- When a cron expression matches, the system **sends `taskDescription` as a message** to the `targetAgent`
+- If the target agent is **offline**, the system **auto-starts** it before delivering the task
+- Tasks are stored per-team in `~/.crewly/teams/{teamId}/cron-tasks.json`
+- Supports **IANA timezones** (e.g., `Asia/Shanghai`, `America/New_York`, `UTC`)
+
+#### Cron Skills (use these!)
+
+```bash
+# Create a recurring task
+bash {{ORCHESTRATOR_SKILLS_PATH}}/create-cron/execute.sh '{"cronExpression":"0 9 * * 1-5","timezone":"Asia/Shanghai","targetAgent":"agent-session","targetTeamId":"team-uuid","taskDescription":"Generate daily standup report"}'
+
+# List all cron tasks (optionally filter by agent or enabled)
+bash {{ORCHESTRATOR_SKILLS_PATH}}/list-cron/execute.sh
+bash {{ORCHESTRATOR_SKILLS_PATH}}/list-cron/execute.sh '{"targetAgent":"agent-session"}'
+
+# Update a cron task (change schedule, description, or enable/disable)
+bash {{ORCHESTRATOR_SKILLS_PATH}}/update-cron/execute.sh '{"id":"cron-xxx","cronExpression":"0 10 * * *"}'
+bash {{ORCHESTRATOR_SKILLS_PATH}}/update-cron/execute.sh '{"id":"cron-xxx","enabled":false}'
+
+# Delete a cron task permanently
+bash {{ORCHESTRATOR_SKILLS_PATH}}/cancel-cron/execute.sh '{"id":"cron-xxx"}'
+```
+
+#### API Endpoints (for reference)
+| Method | Endpoint | Description |
+|--------|----------|-------------|
+| POST | `/api/cron-tasks` | Create a cron task |
+| GET | `/api/cron-tasks` | List all cron tasks (supports `?targetAgent=` and `?enabled=` filters) |
+| PATCH | `/api/cron-tasks/:id` | Update a cron task |
+| DELETE | `/api/cron-tasks/:id` | Delete a cron task |
+
+#### Required Fields for Creation
+| Field | Description |
+|-------|-------------|
+| `cronExpression` | 5-field cron (minute hour day month weekday) |
+| `timezone` | IANA timezone string (default: UTC) |
+| `targetAgent` | Session name of the target agent |
+| `targetTeamId` | Team ID the agent belongs to |
+| `taskDescription` | Message/task sent to the agent on each trigger |
+| `createdBy` | Who created it (use `"orchestrator"`) |
+
+#### Common Cron Expressions
+| Expression | Meaning |
+|------------|---------|
+| `0 9 * * 1-5` | 9:00 AM weekdays |
+| `*/30 * * * *` | Every 30 minutes |
+| `0 0 * * 0` | Midnight every Sunday |
+| `0 8,17 * * *` | 8 AM and 5 PM daily |
+| `0 9 * * 1` | 9 AM every Monday |
+
+⚠️ **IMPORTANT:** Never use Claude's built-in `CronCreate` tool for recurring tasks — those are session-scoped and lost on restart. Always use the Crewly cron skills above.
+
 ### Self-Improvement
 You have access to the `self_improve` tool to safely modify the Crewly codebase:
 - Always create a plan before making changes

package/config/roles/architect/prompt.md

@@ -96,6 +96,15 @@ You have bash skills that let you store and retrieve knowledge that persists acr
 4. **Store project knowledge with `scope: project`** so other agents can benefit
 5. **Store personal knowledge with `scope: agent`** for role-specific learnings
 
+## Recurring Tasks (Cron System)
+
+Your team has a built-in cron system. The orchestrator or user can schedule recurring tasks that are automatically sent to you on a schedule (e.g., daily reports, weekly checks). When a cron task fires:
+
+- You will receive the task description as a normal message in your terminal
+- Treat it like any other task from the orchestrator — execute it and report results
+- If you were offline, Crewly auto-started you to deliver the task
+
+You do not need to manage cron tasks yourself — the orchestrator handles creation and scheduling. If you need a recurring task set up, ask the orchestrator.
 After checking in, just say "Ready for tasks" and wait for me to send you work.
 
 ## Error Learning Protocol

package/config/roles/backend-developer/prompt.md

@@ -113,6 +113,15 @@ You have bash skills that let you store and retrieve knowledge that persists acr
   bash {{AGENT_SKILLS_PATH}}/core/record-learning/execute.sh '{"agentId":"{{SESSION_NAME}}","agentRole":"{{ROLE}}","projectPath":"{{PROJECT_PATH}}","learning":"Current progress: [what was done]. Remaining: [what is left]. Key findings: [important notes]"}'
   ```
 
+## Recurring Tasks (Cron System)
+
+Your team has a built-in cron system. The orchestrator or user can schedule recurring tasks that are automatically sent to you on a schedule (e.g., daily reports, weekly checks). When a cron task fires:
+
+- You will receive the task description as a normal message in your terminal
+- Treat it like any other task from the orchestrator — execute it and report results
+- If you were offline, Crewly auto-started you to deliver the task
+
+You do not need to manage cron tasks yourself — the orchestrator handles creation and scheduling. If you need a recurring task set up, ask the orchestrator.
 After checking in, just say "Ready for tasks" and wait for me to send you work.
 
 ## Error Learning Protocol

package/config/roles/content-strategist/prompt.md

@@ -71,3 +71,13 @@ When you encounter an error and successfully resolve it:
 1. Immediately run `record-learning` with the exact error, fix, and environment context.
 2. If the fix is broadly reusable, store it with `remember` at project scope so other agents inherit it.
 3. Do not finish the task without recording at least one actionable learning when debugging occurred.
+
+## Recurring Tasks (Cron System)
+
+Your team has a built-in cron system. The orchestrator or user can schedule recurring tasks that are automatically sent to you on a schedule (e.g., daily reports, weekly checks). When a cron task fires:
+
+- You will receive the task description as a normal message in your terminal
+- Treat it like any other task from the orchestrator — execute it and report results
+- If you were offline, Crewly auto-started you to deliver the task
+
+You do not need to manage cron tasks yourself — the orchestrator handles creation and scheduling. If you need a recurring task set up, ask the orchestrator.

package/config/roles/designer/prompt.md

@@ -95,6 +95,15 @@ You have bash skills that let you store and retrieve knowledge that persists acr
 4. **Store project knowledge with `scope: project`** so other agents can benefit
 5. **Store personal knowledge with `scope: agent`** for role-specific learnings
 
+## Recurring Tasks (Cron System)
+
+Your team has a built-in cron system. The orchestrator or user can schedule recurring tasks that are automatically sent to you on a schedule (e.g., daily reports, weekly checks). When a cron task fires:
+
+- You will receive the task description as a normal message in your terminal
+- Treat it like any other task from the orchestrator — execute it and report results
+- If you were offline, Crewly auto-started you to deliver the task
+
+You do not need to manage cron tasks yourself — the orchestrator handles creation and scheduling. If you need a recurring task set up, ask the orchestrator.
 After checking in, just say "Ready for tasks" and wait for me to send you work.
 
 ## Error Learning Protocol

package/config/roles/developer/prompt.md

@@ -124,6 +124,15 @@ You have bash skills that let you store and retrieve knowledge that persists acr
   bash {{AGENT_SKILLS_PATH}}/core/record-learning/execute.sh '{"agentId":"{{SESSION_NAME}}","agentRole":"{{ROLE}}","projectPath":"{{PROJECT_PATH}}","learning":"Current progress: [what was done]. Remaining: [what is left]. Key findings: [important notes]"}'
   ```
 
+## Recurring Tasks (Cron System)
+
+Your team has a built-in cron system. The orchestrator or user can schedule recurring tasks that are automatically sent to you on a schedule (e.g., daily reports, weekly checks). When a cron task fires:
+
+- You will receive the task description as a normal message in your terminal
+- Treat it like any other task from the orchestrator — execute it and report results
+- If you were offline, Crewly auto-started you to deliver the task
+
+You do not need to manage cron tasks yourself — the orchestrator handles creation and scheduling. If you need a recurring task set up, ask the orchestrator.
 After checking in, just say "Ready for tasks" and wait for me to send you work.
 
 ## Error Learning Protocol

package/config/roles/frontend-developer/prompt.md

@@ -113,6 +113,15 @@ You have bash skills that let you store and retrieve knowledge that persists acr
   bash {{AGENT_SKILLS_PATH}}/core/record-learning/execute.sh '{"agentId":"{{SESSION_NAME}}","agentRole":"{{ROLE}}","projectPath":"{{PROJECT_PATH}}","learning":"Current progress: [what was done]. Remaining: [what is left]. Key findings: [important notes]"}'
   ```
 
+## Recurring Tasks (Cron System)
+
+Your team has a built-in cron system. The orchestrator or user can schedule recurring tasks that are automatically sent to you on a schedule (e.g., daily reports, weekly checks). When a cron task fires:
+
+- You will receive the task description as a normal message in your terminal
+- Treat it like any other task from the orchestrator — execute it and report results
+- If you were offline, Crewly auto-started you to deliver the task
+
+You do not need to manage cron tasks yourself — the orchestrator handles creation and scheduling. If you need a recurring task set up, ask the orchestrator.
 After checking in, just say "Ready for tasks" and wait for me to send you work.
 
 ## Error Learning Protocol

package/config/roles/fullstack-dev/prompt.md

@@ -113,6 +113,15 @@ You have bash skills that let you store and retrieve knowledge that persists acr
   bash {{AGENT_SKILLS_PATH}}/core/record-learning/execute.sh '{"agentId":"{{SESSION_NAME}}","agentRole":"{{ROLE}}","projectPath":"{{PROJECT_PATH}}","learning":"Current progress: [what was done]. Remaining: [what is left]. Key findings: [important notes]"}'
   ```
 
+## Recurring Tasks (Cron System)
+
+Your team has a built-in cron system. The orchestrator or user can schedule recurring tasks that are automatically sent to you on a schedule (e.g., daily reports, weekly checks). When a cron task fires:
+
+- You will receive the task description as a normal message in your terminal
+- Treat it like any other task from the orchestrator — execute it and report results
+- If you were offline, Crewly auto-started you to deliver the task
+
+You do not need to manage cron tasks yourself — the orchestrator handles creation and scheduling. If you need a recurring task set up, ask the orchestrator.
 After checking in, just say "Ready for tasks" and wait for me to send you work.
 
 ## Error Learning Protocol

package/config/roles/generalist/prompt.md

@@ -103,4 +103,13 @@ You have bash skills that let you store and retrieve knowledge that persists acr
 5. **Store personal knowledge with `scope: agent`** for role-specific learnings
 6. **Store cross-agent user preferences with `scope: project` + `category: user_preference`**
 
+## Recurring Tasks (Cron System)
+
+Your team has a built-in cron system. The orchestrator or user can schedule recurring tasks that are automatically sent to you on a schedule (e.g., daily reports, weekly checks). When a cron task fires:
+
+- You will receive the task description as a normal message in your terminal
+- Treat it like any other task from the orchestrator — execute it and report results
+- If you were offline, Crewly auto-started you to deliver the task
+
+You do not need to manage cron tasks yourself — the orchestrator handles creation and scheduling. If you need a recurring task set up, ask the orchestrator.
 After checking in, just say "Ready for tasks" and wait for me to send you work.

package/config/roles/ops/prompt.md

@@ -127,6 +127,15 @@ You have bash skills that let you store and retrieve knowledge that persists acr
   bash {{AGENT_SKILLS_PATH}}/core/record-learning/execute.sh '{"agentId":"{{SESSION_NAME}}","agentRole":"{{ROLE}}","projectPath":"{{PROJECT_PATH}}","learning":"Current progress: [what was done]. Remaining: [what is left]. Key findings: [important notes]"}'
   ```
 
+## Recurring Tasks (Cron System)
+
+Your team has a built-in cron system. The orchestrator or user can schedule recurring tasks that are automatically sent to you on a schedule (e.g., daily reports, weekly checks). When a cron task fires:
+
+- You will receive the task description as a normal message in your terminal
+- Treat it like any other task from the orchestrator — execute it and report results
+- If you were offline, Crewly auto-started you to deliver the task
+
+You do not need to manage cron tasks yourself — the orchestrator handles creation and scheduling. If you need a recurring task set up, ask the orchestrator.
 After checking in, just say "Ready for tasks" and wait for me to send you work.
 
 ## Error Learning Protocol

package/config/roles/product-manager/prompt.md

@@ -100,6 +100,15 @@ You have bash skills that let you store and retrieve knowledge that persists acr
 
 ## Startup Protocol (#143)
 
+## Recurring Tasks (Cron System)
+
+Your team has a built-in cron system. The orchestrator or user can schedule recurring tasks that are automatically sent to you on a schedule (e.g., daily reports, weekly checks). When a cron task fires:
+
+- You will receive the task description as a normal message in your terminal
+- Treat it like any other task from the orchestrator — execute it and report results
+- If you were offline, Crewly auto-started you to deliver the task
+
+You do not need to manage cron tasks yourself — the orchestrator handles creation and scheduling. If you need a recurring task set up, ask the orchestrator.
 After checking in, **before** saying "Ready for tasks", perform these startup steps:
 
 1. **Recall active goals** — Run `recall` with context "active goals, roadmap, pending work, sprint status" to load previous knowledge about what needs to be done.

package/config/roles/qa/prompt.md

@@ -113,6 +113,15 @@ You have bash skills that let you store and retrieve knowledge that persists acr
   bash {{AGENT_SKILLS_PATH}}/core/record-learning/execute.sh '{"agentId":"{{SESSION_NAME}}","agentRole":"{{ROLE}}","projectPath":"{{PROJECT_PATH}}","learning":"Current progress: [what was done]. Remaining: [what is left]. Key findings: [important notes]"}'
   ```
 
+## Recurring Tasks (Cron System)
+
+Your team has a built-in cron system. The orchestrator or user can schedule recurring tasks that are automatically sent to you on a schedule (e.g., daily reports, weekly checks). When a cron task fires:
+
+- You will receive the task description as a normal message in your terminal
+- Treat it like any other task from the orchestrator — execute it and report results
+- If you were offline, Crewly auto-started you to deliver the task
+
+You do not need to manage cron tasks yourself — the orchestrator handles creation and scheduling. If you need a recurring task set up, ask the orchestrator.
 After checking in, just say "Ready for tasks" and wait for me to send you work.
 
 ## Error Learning Protocol

package/config/roles/qa-engineer/prompt.md

@@ -113,6 +113,15 @@ You have bash skills that let you store and retrieve knowledge that persists acr
   bash {{AGENT_SKILLS_PATH}}/core/record-learning/execute.sh '{"agentId":"{{SESSION_NAME}}","agentRole":"{{ROLE}}","projectPath":"{{PROJECT_PATH}}","learning":"Current progress: [what was done]. Remaining: [what is left]. Key findings: [important notes]"}'
   ```
 
+## Recurring Tasks (Cron System)
+
+Your team has a built-in cron system. The orchestrator or user can schedule recurring tasks that are automatically sent to you on a schedule (e.g., daily reports, weekly checks). When a cron task fires:
+
+- You will receive the task description as a normal message in your terminal
+- Treat it like any other task from the orchestrator — execute it and report results
+- If you were offline, Crewly auto-started you to deliver the task
+
+You do not need to manage cron tasks yourself — the orchestrator handles creation and scheduling. If you need a recurring task set up, ask the orchestrator.
 After checking in, just say "Ready for tasks" and wait for me to send you work.
 
 ## Error Learning Protocol

package/config/roles/researcher/prompt.md

@@ -96,6 +96,15 @@ You have bash skills that let you store and retrieve knowledge that persists acr
 4. **Store project knowledge with `scope: project`** so other agents can benefit
 5. **Store personal knowledge with `scope: agent`** for role-specific learnings
 
+## Recurring Tasks (Cron System)
+
+Your team has a built-in cron system. The orchestrator or user can schedule recurring tasks that are automatically sent to you on a schedule (e.g., daily reports, weekly checks). When a cron task fires:
+
+- You will receive the task description as a normal message in your terminal
+- Treat it like any other task from the orchestrator — execute it and report results
+- If you were offline, Crewly auto-started you to deliver the task
+
+You do not need to manage cron tasks yourself — the orchestrator handles creation and scheduling. If you need a recurring task set up, ask the orchestrator.
 After checking in, just say "Ready for tasks" and wait for me to send you work.
 
 ## Error Learning Protocol

package/config/roles/sales/prompt.md

@@ -95,6 +95,15 @@ You have bash skills that let you store and retrieve knowledge that persists acr
 4. **Store project knowledge with `scope: project`** so other agents can benefit
 5. **Store personal knowledge with `scope: agent`** for role-specific learnings
 
+## Recurring Tasks (Cron System)
+
+Your team has a built-in cron system. The orchestrator or user can schedule recurring tasks that are automatically sent to you on a schedule (e.g., daily reports, weekly checks). When a cron task fires:
+
+- You will receive the task description as a normal message in your terminal
+- Treat it like any other task from the orchestrator — execute it and report results
+- If you were offline, Crewly auto-started you to deliver the task
+
+You do not need to manage cron tasks yourself — the orchestrator handles creation and scheduling. If you need a recurring task set up, ask the orchestrator.
 After checking in, just say "Ready for tasks" and wait for me to send you work.
 
 ## Error Learning Protocol

package/config/roles/support/prompt.md

@@ -95,6 +95,15 @@ You have bash skills that let you store and retrieve knowledge that persists acr
 4. **Store project knowledge with `scope: project`** so other agents can benefit
 5. **Store personal knowledge with `scope: agent`** for role-specific learnings
 
+## Recurring Tasks (Cron System)
+
+Your team has a built-in cron system. The orchestrator or user can schedule recurring tasks that are automatically sent to you on a schedule (e.g., daily reports, weekly checks). When a cron task fires:
+
+- You will receive the task description as a normal message in your terminal
+- Treat it like any other task from the orchestrator — execute it and report results
+- If you were offline, Crewly auto-started you to deliver the task
+
+You do not need to manage cron tasks yourself — the orchestrator handles creation and scheduling. If you need a recurring task set up, ask the orchestrator.
 After checking in, just say "Ready for tasks" and wait for me to send you work.
 
 ## Error Learning Protocol

package/config/roles/team-leader/prompt.md

@@ -18,6 +18,8 @@ You are a **Team Leader** (hierarchyLevel=1) responsible for managing a sub-team
 
 **Core identity**: You are a manager, not an individual contributor. Your first reaction should be "Who is best suited for this?" — not "How do I write this code/content myself?" Delegate 90% of execution tasks to your workers. Only handle complex coordination yourself.
 
+**Worker activation**: Before delegating a task, check worker status via `get-team-status`. If the target worker is **inactive**, use `start-agent` to activate them first, then wait for them to report ready before delegating. Never skip delegation just because workers are offline — wake them up.
+
 **Hierarchy position**: You report directly to the Orchestrator and manage all workers whose `parentMemberId` points to you.
 
 ## Your Workers
@@ -185,4 +187,13 @@ Please run the register-self skill to let the team dashboard know you're availab
 bash {{AGENT_SKILLS_PATH}}/core/register-self/execute.sh '{"role":"{{ROLE}}","sessionName":"{{SESSION_NAME}}"}'
 ```
 
+## Recurring Tasks (Cron System)
+
+Your team has a built-in cron system. The orchestrator or user can schedule recurring tasks that are automatically sent to you on a schedule (e.g., daily reports, weekly checks). When a cron task fires:
+
+- You will receive the task description as a normal message in your terminal
+- Treat it like any other task from the orchestrator — execute it and report results
+- If you were offline, Crewly auto-started you to deliver the task
+
+You do not need to manage cron tasks yourself — the orchestrator handles creation and scheduling. If you need a recurring task set up, ask the orchestrator.
 After checking in, say "Ready for tasks" and wait for the Orchestrator to send you work.

package/config/roles/tpm/prompt.md

@@ -96,6 +96,15 @@ You have bash skills that let you store and retrieve knowledge that persists acr
 4. **Store project knowledge with `scope: project`** so other agents can benefit
 5. **Store personal knowledge with `scope: agent`** for role-specific learnings
 
+## Recurring Tasks (Cron System)
+
+Your team has a built-in cron system. The orchestrator or user can schedule recurring tasks that are automatically sent to you on a schedule (e.g., daily reports, weekly checks). When a cron task fires:
+
+- You will receive the task description as a normal message in your terminal
+- Treat it like any other task from the orchestrator — execute it and report results
+- If you were offline, Crewly auto-started you to deliver the task
+
+You do not need to manage cron tasks yourself — the orchestrator handles creation and scheduling. If you need a recurring task set up, ask the orchestrator.
 After checking in, just say "Ready for tasks" and wait for me to send you work.
 
 ## Error Learning Protocol

package/config/roles/ux-designer/prompt.md

@@ -97,6 +97,15 @@ You have bash skills that let you store and retrieve knowledge that persists acr
 4. **Store project knowledge with `scope: project`** so other agents can benefit
 5. **Store personal knowledge with `scope: agent`** for role-specific learnings
 
+## Recurring Tasks (Cron System)
+
+Your team has a built-in cron system. The orchestrator or user can schedule recurring tasks that are automatically sent to you on a schedule (e.g., daily reports, weekly checks). When a cron task fires:
+
+- You will receive the task description as a normal message in your terminal
+- Treat it like any other task from the orchestrator — execute it and report results
+- If you were offline, Crewly auto-started you to deliver the task
+
+You do not need to manage cron tasks yourself — the orchestrator handles creation and scheduling. If you need a recurring task set up, ask the orchestrator.
 After checking in, just say "Ready for tasks" and wait for me to send you work.
 
 ## Error Learning Protocol

package/config/skills/agent/core/block-task/execute.sh

@@ -16,10 +16,12 @@ require_param "reason" "$REASON"
 
 BODY=$(jq -n \
   --arg absoluteTaskPath "$ABSOLUTE_TASK_PATH" \
+  --arg taskPath "$ABSOLUTE_TASK_PATH" \
   --arg reason "$REASON" \
+  --arg blockReason "$REASON" \
   --arg questions "$QUESTIONS" \
   --arg urgency "$URGENCY" \
-  '{absoluteTaskPath: $absoluteTaskPath, reason: $reason} +
+  '{absoluteTaskPath: $absoluteTaskPath, taskPath: $taskPath, reason: $reason, blockReason: $blockReason} +
    (if $questions != "" then {questions: $questions} else {} end) +
    (if $urgency != "" then {urgency: $urgency} else {} end)')
 

package/config/skills/agent/core/pipe-to-sink/execute.sh

@@ -0,0 +1,41 @@
+#!/usr/bin/env bash
+# pipe-to-sink: Route structured data to a registered sink via the Data API.
+#
+# Input (JSON):
+#   {
+#     "sinkId": "steve-inputs",
+#     "data": { "title": "...", "source": "slack", ... },
+#     "projectPath": "/path/to/project"    (optional — falls back to CREWLY_PROJECT_PATH)
+#   }
+#
+# Calls: POST /api/v2/data/sinks/:sinkId
+# Returns: JSON with the created DataObject ID on success, or error details.
+set -euo pipefail
+SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
+source "${SCRIPT_DIR}/../../_common/lib.sh"
+
+INPUT=$(read_json_input "${1:-}")
+[ -z "$INPUT" ] && error_exit "Usage: execute.sh '{\"sinkId\":\"...\",\"data\":{...}}' or echo '{...}' | execute.sh"
+
+SINK_ID=$(printf '%s' "$INPUT" | jq -r '.sinkId // empty')
+require_param "sinkId" "$SINK_ID"
+
+DATA=$(printf '%s' "$INPUT" | jq '.data // empty')
+if [ "$DATA" = "null" ] || [ "$DATA" = "" ] || [ "$DATA" = "empty" ]; then
+  error_exit "Missing required parameter: data"
+fi
+
+# Resolve projectPath: explicit > env var > omit
+PROJECT_PATH=$(printf '%s' "$INPUT" | jq -r '.projectPath // empty')
+if [ -z "$PROJECT_PATH" ] && [ -n "${CREWLY_PROJECT_PATH:-}" ]; then
+  PROJECT_PATH="$CREWLY_PROJECT_PATH"
+fi
+
+# Build request body
+BODY=$(jq -n \
+  --argjson data "$DATA" \
+  --arg projectPath "$PROJECT_PATH" \
+  '{data: $data} + (if $projectPath != "" then {projectPath: $projectPath} else {} end)')
+
+# POST to the data API
+api_call POST "/v2/data/sinks/${SINK_ID}" "$BODY"

package/config/skills/agent/core/read-task/execute.sh

@@ -10,6 +10,8 @@ INPUT=$(read_json_input "${1:-}")
 ABSOLUTE_TASK_PATH=$(printf '%s' "$INPUT" | jq -r '.absoluteTaskPath // empty')
 require_param "absoluteTaskPath" "$ABSOLUTE_TASK_PATH"
 
-BODY=$(jq -n --arg absoluteTaskPath "$ABSOLUTE_TASK_PATH" '{absoluteTaskPath: $absoluteTaskPath}')
+BODY=$(jq -n --arg absoluteTaskPath "$ABSOLUTE_TASK_PATH" \
+  --arg taskPath "$ABSOLUTE_TASK_PATH" \
+  '{absoluteTaskPath: $absoluteTaskPath, taskPath: $taskPath}')
 
 api_call POST "/task-management/read-task" "$BODY"

package/config/skills/agent/core/report-progress/execute.sh

@@ -5,7 +5,7 @@ SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
 source "${SCRIPT_DIR}/../../_common/lib.sh"
 
 INPUT=$(read_json_input "${1:-}")
-[ -z "$INPUT" ] && error_exit "Usage: execute.sh '{\"sessionName\":\"dev-1\",\"progress\":50,\"current\":\"Implementing tests\"}'"
+[ -z "$INPUT" ] && error_exit "Usage: execute.sh '{\"sessionName\":\"dev-1\",\"progress\":50,\"current\":\"Implementing tests\",\"taskGroup\":\"delegated\"}'"
 
 SESSION_NAME=$(printf '%s' "$INPUT" | jq -r '.sessionName // empty')
 PROGRESS=$(printf '%s' "$INPUT" | jq -r '.progress // empty')
@@ -20,6 +20,8 @@ BODY=$(printf '%s' "$INPUT" | jq '{
   progress: (.progress | tonumber),
   current: .current
 } +
+  (if .projectPath then {projectPath: .projectPath} else {} end) +
+  (if .taskGroup then {taskGroup: .taskGroup} else {} end) +
   (if .completed then {completed: .completed} else {} end) +
   (if .nextSteps then {nextSteps: .nextSteps} else {} end) +
   (if .blockers then {blockers: .blockers} else {} end) +

package/config/skills/agent/screenshot-compare/SKILL.md

@@ -0,0 +1,75 @@
+---
+name: screenshot-compare
+description: Compare two screenshots (e.g., iOS reference vs Web implementation) using Gemini Vision API. Returns a structured diff report with categorized issues, severity levels, and CSS fix suggestions. Use when verifying UI parity, checking design implementation accuracy, or auditing visual consistency across platforms.
+category: qa
+assignableRoles:
+  - "*"
+version: "1.0.0"
+tags:
+  - screenshot
+  - visual-qa
+  - comparison
+  - gemini
+  - parity
+---
+
+# Screenshot Compare
+
+Compare two UI screenshots using Gemini Vision API and get a structured diff report.
+
+## Actions
+
+| Parameter | Required | Description |
+|-----------|----------|-------------|
+| `reference` | Yes | Path to the reference screenshot (source of truth, e.g., iOS app) |
+| `target` | Yes | Path to the target screenshot (implementation to verify, e.g., web app) |
+| `focus` | No | Comma-separated focus areas: `icons,layout,colors,text,images,spacing` (default: all) |
+| `context` | No | Additional context (e.g., "This is the settings page", "Dark mode") |
+
+## Usage
+
+```bash
+# Basic comparison
+bash execute.sh '{"reference":"/path/to/ios-screenshot.png","target":"/path/to/web-screenshot.png"}'
+
+# Focused comparison (only icons and colors)
+bash execute.sh '{"reference":"ref.png","target":"web.png","focus":"icons,colors"}'
+
+# With context
+bash execute.sh '{"reference":"ios-home.png","target":"web-home.png","context":"Home screen, light mode"}'
+```
+
+## Output Format
+
+Returns JSON with:
+- `matchScore` — 0-100 overall similarity score
+- `totalIssues` — count of differences found
+- `issues[]` — array of categorized differences:
+  - `type` — icon_missing, layout_shift, color_mismatch, text_mismatch, font_mismatch, spacing, etc.
+  - `severity` — critical, major, minor
+  - `element` — which UI element is affected
+  - `suggestion` — specific CSS/code fix
+
+## Issue Types
+
+| Type | Description |
+|------|-------------|
+| `icon_missing` | Icon present in reference but missing in target |
+| `icon_wrong` | Icon exists but is the wrong icon |
+| `layout_shift` | Elements positioned differently |
+| `spacing` | Padding/margin differences |
+| `color_mismatch` | Colors don't match between reference and target |
+| `text_mismatch` | Text content differs |
+| `font_mismatch` | Font family, size, or weight differs |
+| `image_missing` | Image present in reference but missing/broken in target |
+| `image_wrong` | Image exists but is different |
+| `border_radius` | Rounded corners differ |
+| `shadow` | Box shadow differences |
+| `alignment` | Vertical/horizontal alignment issues |
+| `responsive` | Layout doesn't adapt correctly |
+
+## Prerequisites
+
+- `GEMINI_API_KEY` environment variable or configured in Settings > API Keys
+- Both image files must exist and be PNG, JPG, GIF, or WebP
+- Images should be under 4MB each