[WIP] Update deepreport to extract actionable agentic tasks (#6846)

Author: Copilot

.github/workflows/deep-report.lock.yml

@@ -66,6 +66,7 @@ Continuously review and aggregate information from the various reports created a
 3. **Flag interesting activity** - Highlight noteworthy discoveries, improvements, or anomalies
 4. **Detect suspicious patterns** - Identify potential security concerns or concerning behaviors
 5. **Surface exciting developments** - Celebrate wins, improvements, and positive trends
+6. **Extract actionable tasks** - Identify exactly 3 specific, high-impact tasks that can be assigned to agents for quick wins
 
 ## Data Sources
 
@@ -183,6 +184,38 @@ Connect the dots between different data sources:
 2. Identify agents that may be experiencing issues
 3. Find patterns that span multiple report types
 4. Track how identified patterns evolve over time
+5. **Identify improvement opportunities** - Look for:
+   - Duplicate or inefficient patterns that can be consolidated
+   - Missing configurations (caching, error handling, documentation)
+   - High token usage in workflows that could be optimized
+   - Repetitive manual tasks that can be automated
+   - Issues or discussions that need attention (labeling, triage, responses)
+
+### Step 3.5: Extract Actionable Agentic Tasks
+
+**CRITICAL**: Based on your analysis, identify exactly **3 actionable tasks** (quick wins):
+
+1. **Prioritize by impact and effort**: Look for high-impact, low-effort improvements
+2. **Be specific**: Tasks should be concrete with clear success criteria
+3. **Consider agent capabilities**: Tasks should be suitable for AI agent execution
+4. **Base on data**: Use insights from discussions, workflows, and issues
+5. **Focus on quick wins**: Tasks that can be completed quickly (< 4 hours of agent time)
+
+**Common quick win categories:**
+- **Code/Configuration improvements**: Consolidate patterns, add missing configs, optimize settings
+- **Documentation gaps**: Add or update missing documentation
+- **Issue/Discussion triage**: Label, organize, or respond to backlog items
+- **Workflow optimization**: Reduce token usage, improve caching, fix inefficiencies
+- **Cleanup tasks**: Remove duplicates, archive stale items, organize files
+
+**Output format for each task:**
+```
+Task 1: [Clear title]
+- Description: [What needs to be done and why]
+- Expected Impact: [Measurable benefit]
+- Suggested Agent: [Which agent can do this]
+- Estimated Effort: [Quick/Medium/Fast]
+```
 
 ### Step 4: Store Insights in Repo Memory
 
@@ -243,6 +276,31 @@ Based on trend analysis, provide:
 - Suggestions for new agents or capabilities
 - Areas that need more monitoring
 
+### ✅ Actionable Agentic Tasks (Quick Wins)
+
+**IMPORTANT**: Extract exactly **3 actionable tasks** that could be immediately assigned to an AI agent to improve the project. Focus on **quick wins** - tasks that are:
+- **Specific and well-defined** - Clear scope with measurable outcome
+- **Achievable by an agent** - Can be automated or assisted by AI
+- **High impact, low effort** - Maximum benefit with minimal implementation time
+- **Data-driven** - Based on patterns and insights from this analysis
+- **Independent** - Can be completed without blocking dependencies
+
+For each task, provide:
+1. **Task Title** - Clear, action-oriented name (e.g., "Reduce token usage in daily-news workflow")
+2. **Description** - 2-3 sentences explaining what needs to be done and why
+3. **Expected Impact** - What improvement or benefit this will deliver
+4. **Suggested Agent** - Which existing agent could handle this, or suggest "New Agent" if needed
+5. **Estimated Effort** - Quick (< 1 hour), Medium (1-4 hours), or Fast (< 30 min)
+
+**If no actionable tasks are identified** (the project is in excellent shape), explicitly state: "No actionable tasks identified - the project is operating optimally."
+
+**Examples of good actionable tasks:**
+- "Consolidate duplicate error handling patterns in 5 workflow files"
+- "Add missing cache configuration to 3 high-frequency workflows"
+- "Create automated labels for 10 unlabeled issues based on content analysis"
+- "Optimize token usage in verbose agent prompts (identified 4 candidates)"
+- "Add missing documentation for 2 frequently-used MCP tools"
+
 ### 📚 Source Attribution
 
 List all reports and data sources analyzed: