Minnie Improvements Backlog

Running list of todos to improve Minnie's capacity

Last updated: 2026-02-09

---

🚨 Critical Issues (Blocking Functionality)

CRON-001: Gateway Token Mismatch (Timers/Reminders Broken)

Status: BLOCKED
Priority: HIGH
Issue: cron tool returns "gateway token mismatch" error when trying to add jobs
Impact: Cannot set timers, one-shot reminders, or create new cron jobs
Error: gateway closed (1008): unauthorized: gateway token mismatch
Attempted: 2026-02-09 when Quan requested 9-minute timer for Jedi Council meeting
Next Step:

• Research OpenClaw cron tool requirements (gateway auth token parameter?)
• Check if this is a known issue in OpenClaw docs
• May need Quan to provide gateway token or reconfigure OpenClaw

---

📋 Phase 4 Blockers (Prevents Advanced Automation)

REF-001: Empty ZTAG Reference Files

Status: BLOCKED (waiting on Quan)
Priority: HIGH
Files needed:

• reference/ztag/overview.md - Product summary, target customers, revenue model
• reference/ztag/districts.md - Current customers, prospects, past outreach outcomes
• reference/ztag/grant-cycles.md - ELOP/CCLC grant windows, purchasing timing
• reference/ztag/brand-guide.md - Messaging, competitive positioning, value props
• reference/ztag/curriculum.md - Product SKUs, pricing tiers, demo availability
• reference/ztag/zoho-ecosystem.md - Which Zoho apps in use, data flow

Impact: Cannot draft intelligent district outreach, track grant cycles, or provide conference prep context
Blocker identified: 2026-02-09
Estimated time to populate: 2-3 hours (rough notes acceptable)
Next step: Quan to populate during next deep-work block, OR I interview him to fill in

INT-002: Zoho Campaigns Not Activated

Status: PENDING (needs confirmation)
Priority: MEDIUM
Issue: Unknown if Zoho Campaigns is activated in Quan's Zoho account
Impact: Can't set up email sequences for 531-district outreach
Next step:

1. Confirm if Zoho Campaigns is active
2. If not: Activate (takes 5 min, free/cheap add-on)
3. If yes: Import 531 contacts + create sequences

---

🔄 Data Quality & Hygiene

CRM-001: Nurture Stage Cleanup

Status: IDENTIFIED
Priority: MEDIUM
Issue: 53 deals in "Nurture" stage ($710K value), many stale (60-90+ days since last modified)
Impact: Pipeline noise, inaccurate forecasting, wasted follow-up effort
Examples:

• Jefferson County Community Ed - 2 months stale
• St. Labre Indian School - 2.5 months stale
  Action needed:
• Define "stale" criteria (60 days? 90 days?)
• Bulk update stale nurture deals to "Closed Lost" or "On Hold"
• Re-engage top 10 by value if still viable
  Next step: Draft cleanup criteria, get Quan approval, execute

CRM-002: Cross-Reference Closed Wins with Outreach List

Status: READY TO EXECUTE
Priority: MEDIUM
Issue: 531-contact CSV may include existing customers (216 closed wins)
Impact: Risk of re-prospecting existing customers (bad experience)
Action needed:

• Match 216 "Deal Closed" accounts to 531-contact CSV by district name
• Tag existing customers (different sequence: upsell vs cold outreach)
• Identify true "new lookalikes" (never engaged before)
  Next step: Run matching script, report overlap to Quan

---

🛠️ Technical Improvements

TOOL-001: Gmail OAuth Auto-Refresh

Status: COMPLETE ✅
Completed: 2026-02-09
Result: Tokens auto-refresh on 401 errors, no daily re-auth needed

TOOL-002: Multi-Account Gmail Support

Status: COMPLETE ✅
Completed: 2026-02-09
Result: gmail-fetch.py supports --account parameter (ztag/gantom/quan777)

TOOL-003: Smart Email Triage (AI Intelligence)

Status: COMPLETE ✅
Completed: 2026-02-09
Result: AI reads emails with context/tone, learns from corrections via memory/email-patterns.md

INT-001: Zoho CRM Connected

Status: COMPLETE ✅
Completed: 2026-02-09
Result: OAuth working, read access to Leads/Contacts/Accounts/Deals

---

📊 Analytics & Reporting

REPORT-001: Weekly Strategic Digest

Status: DESIGNED (not implemented)
Priority: MEDIUM
Description: Friday 6 PM digest covering:

• What each agent accomplished this week
• Cost breakdown by agent/day
• Grant pipeline status (districts contacted, response rate)
• Competitor/market updates
• Recommended actions for next week
  Blocker: Needs Phase 4 reference files + active workflows to summarize
  Next step: Activate after Phase 4 workflows are running

---

🚀 Future Enhancements (Phase 5+)

RAG-001: ChromaDB Memory Layer

Status: PLANNED (Phase 5, months 3-6)
Description: Vector database for intelligent context retrieval
Prerequisites: Phase 4 must deliver 5+ hours/week value first

AGENT-001: Minnie-Retrieval (Agent 5)

Status: PLANNED (Phase 5)
Description: Dedicated retrieval agent for RAG queries
Prerequisites: ChromaDB operational

---

🔐 Security & Compliance

SEC-001: OpenClaw Config Read-Only

Status: ENFORCED
Priority: CRITICAL
Rule: /home/node/.openclaw/openclaw.json is now chmod 444 (read-only)
Reason: Twice crashed gateway by adding invalid keys (heartbeat.intervalMinutes, model.heartbeat, web.braveSearchApiKey)
Constraint: If config change needed, propose to Quan → he applies via SSH
Backup exists: openclaw.json.known-good

---

📝 Documentation Gaps

DOC-001: ZTAG Sales Dynamics

Status: CAPTURED (not formalized)
Priority: LOW
Content: Embodiment gap (conference vs online leads), district-level targeting, pilot → district-wide strategy
Location: Captured in memory/2026-02-09.md
Action: Consider moving to reference/ztag/sales-strategy.md for permanence

---

💬 Communication & UX

UX-001: Push Back on Ambiguity (Preference Learned)

Status: ACTIVE ✅
Learned: 2026-02-09
Rule: When Quan is unclear/ambiguous, ask clarifying questions instead of assuming
Example: Apollo use case - stopped speculating, asked 5 direct questions

---

📈 Metrics to Track

MET-001: Interruptions Created vs Removed (Weekly)

Status: PLANNED
Description: Track whether Minnie adds or reduces cognitive load
Target: Ratio should favor "removed" (net positive value)
Next step: Define what counts as interruption (notifications? questions? approvals?)

MET-002: Time Saved Per Week

Status: PLANNED
Target: 5+ hours/week saved by end of Phase 4
Measurement: Quan self-report + task completion time estimates

---

🎯 How to Use This Backlog

Adding items:

1. Use format: CATEGORY-###: Title
2. Include: Status, Priority, Issue, Impact, Next step
3. Update "Last updated" date at top

Status values:

• BLOCKED - Cannot proceed (dependency missing)
• PENDING - Waiting on decision/input
• READY TO EXECUTE - Can start immediately
• IN PROGRESS - Currently working on it
• COMPLETE ✅ - Done

Priority:

• CRITICAL - System broken, fix immediately
• HIGH - Blocks other work or major impact
• MEDIUM - Valuable but not urgent
• LOW - Nice to have

Review cadence: Weekly (Fridays during strategic digest time)