Complete Mode Quick Reference

Quick lookup guide for using Complete Problem Solving Mode in AI conversations. Bookmark this page for instant access to templates and toggles.

⚡ Quick Start (30 seconds)

@complete-mode

Task = [One sentence: what you're solving]
Context = [Who/where affected, constraints, tools available]

[Pick ONE domain overlay below]

[Optional: Add toggles]

📋 Domain Overlays (Choose One)

🔧 Software/DevOps

[DONE overlay — Software/DevOps]
- Cloud health: /healthz returns 200 for ≥30min; 0 restarts
- Logs clean: last 2min show no errors above INFO
- Local gates: build + tests + lint = 0 failures
- Zero criticals: Problem Register has 0 P0/P1
- Evidence: health URL, log excerpt, test summary, commit refs

📊 Data/Analytics

[DONE overlay — Data/Analytics]
- Objective met: [state exact metric + threshold]
- Statistical validity: `p<0.05` or CI excludes 0; adequate N
- Reproducibility: notebook + data hash provided
- Zero criticals: Problem Register has 0 P0/P1
- Evidence: plots, stats, code + hash, dashboard link

📝 Research/Writing

[DONE overlay — Research/Writing]
- Claims supported: ≥3 primary sources with quotes
- Counterevidence: major counterargs addressed
- Clarity: exec summary + takeaways + limitations
- Zero criticals: Problem Register has 0 P0/P1
- Evidence: bibliography, quoted snippets, reliability notes

🎨 Product/UX

[DONE overlay — Product/UX]
- Ship/prototype: interactive artifact available
- Usability proof: task success rate ≥target on N users
- Acceptance notes: limitations + next iteration documented
- Zero criticals: Problem Register has 0 P0/P1
- Evidence: prototype link, test metrics, decision log

🎚️ Toggles (Optional Controls)

Scope

scope:runtime — Focus on runtime behavior
scope:config — Focus on configuration
scope:migrations — Focus on database changes
scope:security — Focus on security implications
scope:metrics — Focus on measurement
scope:core-flow — Focus on primary user journey
scope:accessibility — Focus on a11y

Depth

depth:shallow — Quick pass, 1-2 discovery methods
depth:normal — Standard, 2-3 discovery methods (default)
depth:deep — Exhaustive, 4+ discovery methods

Risk Tolerance

risk_tolerance:low — Extensive verification required
risk_tolerance:med — Balanced approach (default)
risk_tolerance:high — Accept faster with higher confidence

Other

max_actions:N — Limit changes per pass to N
strict:on — Refuse DONE without all gates proven
require_passes:N — Force minimum N discovery passes

🔄 The Loop (What Happens)

1. DISCOVER  → Find issues, add to Problem Register
2. PLAN      → Prioritize by dependency/impact
3. EXECUTE   → Make changes, capture exact steps
4. VERIFY-1  → Primary proof (tests, metrics, logs)
5. VERIFY-2  → Independent proof (different method)
6. RE-SCAN   → NEW discovery with different techniques
7. UPDATE    → Mark statuses, refuse DONE if gates unmet

Repeat until all acceptance gates pass

📊 Problem Register Template

| ID   | Sev | Category | Evidence          | Root Cause | Action | Status   | Conf |
|------|-----|----------|-------------------|------------|--------|----------|------|
| P-01 | P0  | Runtime  | 502s in /api/auth | Port bind  | Fix cmd| Resolved | 0.9  |
| P-02 | P1  | Config   | Missing DB_URL    | .env issue | Add var| In Prog  | 0.8  |

Severity Levels:

P0 = Critical (production down, data loss, security)
P1 = High (partial outage, wrong results, broken feature)
P2 = Medium (degraded UX, edge case, tech debt)

🎯 When to Use Complete Mode

✅ USE IT FOR:

Production issues
Unclear root causes
High-stakes changes
Compliance/audit needs
Problems that recur
Complex debugging
Team handoffs

❌ DON'T USE FOR:

Quick prototypes
Brainstorming
Exploratory research
Trivial changes
Time-sensitive emergencies (use after stabilization)

🚫 Anti-Shortcut Rules

If only 1 issue found:
→ Run 2+ more discovery passes

If access blocked:
→ State exact blocker, continue with available data

If AI wants to finish early:
→ Add: strict:on require_passes:3 stop_reason:explain

Generic advice rejected:
→ Every claim needs evidence (logs, metrics, tests, links)

💡 Real-World Examples

Example 1: Deploy Failure

@complete-mode

Task = Fix staging deploy returning 503
Context = Render service, have logs + dashboard access, branch=main

[DONE overlay — Software/DevOps]
- Cloud health: /healthz 200 for ≥30min; 0 restarts
- Logs clean: last 2min no errors
- Local gates: build + tests + lint pass
- Zero criticals
- Evidence: health URL, log excerpt, test output, PR link

scope:start-command depth:deep strict:on

Example 2: A/B Test Analysis

@complete-mode

Task = Validate new checkout flow increased conversion
Context = 2 weeks live, 10K users in test, have Amplitude access

[DONE overlay — Data/Analytics]
- Objective met: conversion >baseline by ≥3pp with `p<0.05`
- Statistical validity: adequate sample, CI excludes 0
- Reproducibility: SQL query + Amplitude screenshot
- Zero criticals
- Evidence: plots, stats, queries, dashboard

depth:deep scope:experiment-design

Example 3: API Documentation

@complete-mode

Task = Document webhook signature verification for partners
Context = HMAC-SHA256, partner docs, need code examples in 3 languages

[DONE overlay — Research/Writing]
- Claims supported: ≥3 sources (RFC, our code, security docs)
- Counterevidence: common mistakes documented
- Clarity: summary + examples + troubleshooting
- Zero criticals
- Evidence: links, code snippets, test cases

scope:citations risk_tolerance:low

🔧 Troubleshooting

Problem: AI finishing too fast
Fix: strict:on require_passes:3

Problem: Too much detail
Fix: depth:shallow max_actions:2

Problem: Wrong domain focus
Fix: Add specific scope:area

Problem: Need to skip a gate
Fix: Document why in Problem Register as P2 with rationale

📏 Success Metrics

Track these to measure impact:

Recurrence rate — % of fixes that break again within 2 weeks
Discovery depth — Avg # issues found per session
Verification coverage — % with ≥2 verification methods
Handoff quality — Can teammate understand solution from registers?

Typical improvements: 60-80% fewer recurrences, 3x more issues found

🔗 Full Templates

Action Log (Per Pass)

### Pass N — Action Log
- Change: [diff/command/decision]
- Before → After: [metric showing improvement]
- Verification (Primary): [direct proof]
- Verification (Independent): [different method]
- New signals found? [yes/no + details]

Final Status

**FINAL STATUS:** DONE ✅ (all gates satisfied)

**WHAT TO WATCH:**
- [Residual risk 1]
- [Residual risk 2]
- [Residual risk 3]

**FOLLOW-UPS:**
- [Technical debt item]
- [Next iteration scope]

🎓 Learning Path

Start simple: Use Software/DevOps overlay on a known bug
Practice registers: Maintain Problem Register for 3 sessions
Vary discovery: Consciously use different methods each pass
Add toggles: Experiment with scope/depth controls
Customize gates: Adapt overlays to your team standards
Chain domains: Try multi-overlay approach on complex problems

Complete Problem Solving Framework — Full conceptual guide
Windsurf Rule Installation — Setup for Cascade
Domain-Specific Guides — Deep dives per domain
Real Examples Library — Annotated conversations

💾 Copy-Paste Ready

Minimal invocation:

@complete-mode
Task = [problem]
Context = [constraints]
[DONE overlay — Domain]
strict:on

With all controls:

@complete-mode

Task = [specific problem in one sentence]
Context = [who/where affected, tools, constraints]

[DONE overlay — Domain]
- [Gate 1]
- [Gate 2]
- [Gate 3]
- Zero criticals
- Evidence pack: [artifacts]

scope:[area] depth:[level] risk_tolerance:[level] max_actions:[n] strict:on

Emergency nudge (mid-conversation):

strict:on require_passes:3 stop_reason:explain

Remember: Complete Mode trades initial speed for comprehensive solutions. Use it when thoroughness matters more than time-to-first-answer.

Print this page or bookmark it for quick reference during AI-assisted problem-solving sessions.

Complete Mode Quick Reference: Cheat Sheet for Systematic Problem Solving