agentleFS
Topaz1618Topaz1618/pawskills/.cursor/rules/engineering-context.mdc

pawskills / rules

Cursor rule0 starsChanged 3 days agoMITView on GitHub

Context, logical completeness, technical decision, and implementation authority constraints for engineering problems.

The file

We split this engineering-context.mdc into the rules it gives an agent

12 rules. About 0 are common, 12 are distinctive, and 0 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈12

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 0

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    When several related technical concepts are involved, explain the relationship between them before discussing the concrete implementation.

    line 22
  2. Distinctive:

    Every engineering analysis must clearly distinguish:

    line 32
  3. Distinctive:

    When proposing any technical recommendation, the following must be answered:

    line 43
  4. Distinctive:

    If the current implementation already satisfies the requirement, recommend keeping it as it is, rather than introducing technology for the …

    line 52
  5. Distinctive:

    Must not add database tables, authentication mechanisms, dependency libraries, interfaces, or other functionality on your own initiative.

    line 58
  6. Distinctive:

    When a potential improvement is discovered, it may be raised, but its relationship to the current task must be stated, together with …

    line 60
  7. Distinctive:

    When an earlier conclusion is found to be incorrect, it must be stated which specific conclusion was wrong, why it was wrong, and what the …

    line 68
  8. Distinctive:

    Do not ask the user whether to continue every few sentences.

    line 72
  9. Distinctive:

    When the information is sufficient, give the complete analysis, the recommended approach, and the implementation boundary directly.

    line 74
  10. Distinctive:

    Before modifying code or the database, output:

    line 80
  11. Distinctive:

    Wait for the user's confirmation before executing.

    line 90
  12. Distinctive:

    All engineering discussions, code analyses, architecture designs, database designs, troubleshooting, and code modifications must comply …

    line 94

≈ “Common” is an estimate: at least 1 in 5 of the 6,873 Cursor rule files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
engineering-context.mdc
---
description: Context, logical completeness, technical decision, and implementation authority constraints for engineering problems.
alwaysApply: true
---

# Context and Logical Completeness Constraints for Engineering Problems

## 1. Understand the Problem Before Proposing a Solution

For any technical discussion, identify first:

- What is the specific problem under discussion?
- What does the user want to solve?
- What implementation does the current project actually use?
- Which facts and decisions have already been established earlier?
- How does the current problem relate to the earlier discussion?

It is forbidden to explain a technical term in isolation, detached from context, or to introduce a new technical approach on your own initiative.

## 2. A Complete Chain of Technical Logic Is Required

When several related technical concepts are involved, explain the relationship between them before discussing the concrete implementation.

For example, when Cookie, Session, Token, and JWT are involved:

First explain their respective responsibilities and their relationship, then analyze the combination the current project actually uses, and finally compare the advantages and disadvantages of the alternative approaches.

Concepts at different layers must not be conflated, and it is not sufficient to merely list terminology.

## 3. Strictly Distinguish Four Kinds of Information

Every engineering analysis must clearly distinguish:

- **[Current state]**: facts verified from the current code, configuration, or database.
- **[Problem]**: a specific defect or limitation in the existing implementation.
- **[Recommendation]**: an improvement proposed on the basis of actual requirements.
- **[To be confirmed]**: missing evidence, or business requirements that are not yet settled.

It is forbidden to present your own inference as a project fact, and it is forbidden to describe your own recommendation as a decision the user has already made.

## 4. A Proposal Must Have a Complete Causal Chain

When proposing any technical recommendation, the following must be answered:

1. Why does the current implementation have a problem?
2. What concrete impact does it cause?
3. Why is this approach recommended?
4. How does this approach solve the problem?
5. What are the costs, risks, or side effects?
6. Is there a simpler alternative?

If the current implementation already satisfies the requirement, recommend keeping it as it is, rather than introducing technology for the sake of introducing technology.

## 5. Never Expand the Scope of the Change

The user discussing a problem does not authorize modifying the whole system.

Must not add database tables, authentication mechanisms, dependency libraries, interfaces, or other functionality on your own initiative.

When a potential improvement is discovered, it may be raised, but its relationship to the current task must be stated, together with whether it constitutes a necessary change.

## 6. Maintain Cross-Turn Context

Code facts already verified, technical choices already made, and decisions the user has already taken must serve as the basis for subsequent discussion.

Unless new evidence is found, previously settled conclusions must not be repeatedly overturned, reopened, or replaced with contradictory proposals.

When an earlier conclusion is found to be incorrect, it must be stated which specific conclusion was wrong, why it was wrong, and what the impact on subsequent proposals is.

## 7. A Technical Discussion Must Reach a Complete Conclusion

Do not ask the user whether to continue every few sentences.

When the information is sufficient, give the complete analysis, the recommended approach, and the implementation boundary directly.

Raise one explicit question only when a key unknown exists that genuinely affects the technical decision.

## 8. Output the Plan Before Starting Implementation

Before modifying code or the database, output:

- The current implementation and the files involved.
- The content of the planned changes.
- The reason for the changes.
- The data structure and interface changes.
- Compatibility and risks.
- The testing approach.
- The parts that will explicitly not be modified.

Wait for the user's confirmation before executing.

The user confirming a specific proposal does not mean that unrelated parts may be refactored along the way.

All engineering discussions, code analyses, architecture designs, database designs, troubleshooting, and code modifications must comply with this rule.

This rule must not be skipped on the grounds that the current task did not trigger a particular Skill.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.