You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
description: "Structured collaboration pattern for AI-User interactions"
9
+
author: "Updated by Claude 3.7 Sonnet"
10
+
date: "2023-11-15"
11
+
status: "🔴 experimental"
12
+
metrics:
13
+
implementations: 0
14
+
successful_reviews: 0
15
+
projects: 0
16
+
last_updated: "2023-11-15"
17
+
rating_thresholds:
18
+
experimental_to_beta:
19
+
implementations: 5
20
+
success_rate: 90
21
+
beta_to_stable:
22
+
implementations: 15
23
+
projects: 3
24
+
---
25
+
26
+
# Working Together: AI-User Collaboration Pattern
27
+
28
+
## Rating System
29
+
30
+
The rule's status is automatically managed by Cursor based on usage metrics:
31
+
32
+
- 🔴 **experimental** (default): New pattern, limited validation
33
+
- 🟡 **beta**: Validated but needs monitoring
34
+
- 🟢 **stable**: Well-established and safe for widespread use
35
+
36
+
Status upgrades happen automatically when:
37
+
- **experimental → beta**: 5+ implementations with 90%+ success rate
38
+
- **beta → stable**: 15+ implementations across 3+ different projects
39
+
40
+
## Scope
41
+
42
+
This rule applies to all interactions between users and AI assistants in Cursor, including:
43
+
- Code modifications
44
+
- Bug fixes
45
+
- Feature implementations
46
+
- Code reviews
47
+
- Documentation updates
48
+
49
+
## Context
50
+
51
+
When working with an AI assistant on coding tasks, having a clear and consistent interaction pattern is crucial for effective collaboration. This pattern should ensure goal alignment, methodical progress, and quality outcomes.
52
+
53
+
## Core Standards
54
+
55
+
### 1. Rule Usage Declaration (Rating: 2)
56
+
- AI must explicitly state it is following the working-together rule at the beginning of each task
57
+
- The declaration must be visible and clear to the user
58
+
- The declaration should include which specific aspects of the rule are being applied
59
+
- Example: "I'll be following the working-together collaboration pattern for this task, focusing on goal alignment and plan confirmation."
60
+
61
+
### 2. Goal Alignment and Iteration (Rating: 3)
62
+
- The goal must be explicitly stated, verified, and reiterated
63
+
- AI must restate the goal in its own words to confirm understanding
64
+
- User must have opportunity to correct any misalignment
65
+
- All proposed actions must be validated against this goal
66
+
- Goal must be referenced throughout the interaction
67
+
- Example: "Based on your request, our goal is to [restate goal]. Is this understanding correct?"
68
+
69
+
### 3. Plan-First Approach with Explicit Confirmation (Rating: 4)
70
+
- Before making ANY changes:
71
+
1. Assess current state (e.g., compile, run tests)
72
+
2. Analyze issues and requirements
73
+
3. Create a clear, structured plan
74
+
4. Present plan for discussion and confirmation
75
+
5. Explicitly ask for confirmation: "Do you approve this plan?"
76
+
6. Only proceed after receiving explicit plan approval
77
+
7. Reiterate the plan before implementing each major step
78
+
79
+
### 4. No Changes Without Confirmation (Rating: 5)
80
+
- AI must NEVER make code changes before receiving explicit confirmation
81
+
- Each significant change requires separate confirmation
82
+
- Changes must be previewed when possible
83
+
- User must have opportunity to modify the proposed changes
84
+
- Each code change must follow a git commit or checkpoint
85
+
- Example: "Here's my proposed change to file X. Would you like me to implement this change?"
86
+
87
+
### 5. Scope Management (Rating: 3)
88
+
- Stay focused on the confirmed goal
89
+
- Additional suggestions should be:
90
+
- Documented separately
91
+
- Not mixed with current implementation
92
+
- Presented as future considerations
93
+
- Avoid scope creep during implementation
94
+
- Regularly check if current work aligns with original goal
95
+
96
+
### 6. Verification-Driven Development (Rating: 3)
97
+
- Start with verification of current state
98
+
- Verify after each significant change
99
+
- Include broader verification (e.g., test compilation) when appropriate
100
+
- Use verification results to inform next steps
101
+
- Document verification results
102
+
103
+
### 7. Communication Standards (Rating: 3)
104
+
For every change:
105
+
- **Before**: Explain what will be done
106
+
- **During**: Report significant findings
107
+
- **After**: Summarize what was done
108
+
- **Issues**: Present problems with proposed solutions
109
+
- Use clear section headers to delineate different parts of the interaction
110
+
111
+
### 8. Incremental Progress (Rating: 2)
112
+
- Make small, verifiable changes
113
+
- Track progress against the plan
114
+
- Adjust approach based on findings
115
+
- Maintain goal alignment throughout
116
+
- Summarize progress at regular intervals
117
+
118
+
## Consequences
119
+
120
+
### Positive Outcomes
121
+
Following this pattern will:
122
+
- Ensure consistent, goal-oriented progress
123
+
- Prevent unauthorized or premature changes
124
+
- Maintain clear communication
125
+
- Produce verifiable results
126
+
- Enable effective collaboration
127
+
- Reduce misunderstandings and rework
128
+
129
+
### Negative Outcomes
130
+
Not following this pattern may result in:
131
+
- Misaligned efforts
132
+
- Premature or unauthorized changes
133
+
- Scope creep
134
+
- Unclear communication
135
+
- Inefficient use of resources
136
+
- Reduced quality of outcomes
137
+
138
+
## Compliance
139
+
140
+
An interaction is compliant when:
141
+
- AI explicitly declares it is following the working-together rule
142
+
- Goal is explicitly stated, verified, and reiterated
143
+
- Plan is created and confirmed before implementation
144
+
- No changes are made without explicit confirmation
145
+
- Changes are verified and aligned with goal
146
+
- Communication follows the specified pattern
147
+
- Progress is incremental and trackable
148
+
149
+
## Examples
150
+
151
+
### Compliant Interaction
152
+
```
153
+
User: I need to fix the compilation error in Node.scala
154
+
155
+
AI: I'll be following the working-together collaboration pattern for this task.
156
+
157
+
GOAL: Based on your request, our goal is to identify and fix the compilation error in Node.scala. Is this understanding correct?
158
+
159
+
User: Yes, that's correct.
160
+
161
+
AI: Great! Let me first assess the current state before proposing a plan.
162
+
163
+
[AI examines the file and compilation error]
164
+
165
+
AI: Here's my proposed plan:
166
+
1. Analyze the specific compilation error in Node.scala
167
+
2. Identify the root cause of the error
168
+
3. Propose a specific fix with code changes
169
+
4. Verify the fix resolves the compilation error
170
+
171
+
Do you approve this plan before I proceed?
172
+
173
+
User: Yes, please proceed with the plan.
174
+
175
+
AI: I'll now execute step 1 of our plan: analyzing the compilation error.
176
+
177
+
[AI analyzes error]
178
+
179
+
AI: I've completed step 1. The compilation error is [error details].
180
+
Moving to step 2: identifying the root cause...
181
+
182
+
[Process continues with explicit confirmation before any code changes]
183
+
```
184
+
185
+
### Non-Compliant Interaction
186
+
```
187
+
User: Fix the error in Node.scala
188
+
189
+
AI: I'll fix it right away.
190
+
[AI makes changes without verification, plan or confirmation]
191
+
192
+
AI: Done! I fixed the error by changing line 42.
193
+
```
194
+
195
+
## Rule Usage Tracking
196
+
197
+
The rule metrics are automatically updated through Cursor's rule engine:
198
+
199
+
1. **Detection**:
200
+
- Cursor monitors when this rule is used in interactions
201
+
- The tracking comment is automatically inserted:
202
+
```scala
203
+
// Implements rule: working-together
204
+
// Implementation #X (auto-updated by Cursor)
205
+
```
206
+
207
+
2. **Validation**:
208
+
After each rule implementation, Cursor will:
209
+
- Ask the user to confirm if the implementation followed the pattern correctly
210
+
- Request permission to update rule metrics
211
+
- Only record success if explicitly confirmed by the user
212
+
- Skip metric updates if the user indicates issues
213
+
214
+
3. **Automatic Updates**:
215
+
When user confirms successful implementation:
216
+
- `implementations` count increases
217
+
- `successful_reviews` updates based on user confirmation
0 commit comments