- Home
- Skills
- Code Quality & Review
- Sofware Engineering Rulebook
Works with the AI Agents you already use
Sofware Engineering Rulebook
Review code reliably using a rulebook anchored in software engineering litterature.
$1.99
Sofware Engineering Rulebook
Example session with this skill installed
User Request
"Review this PR for logic errors. The calculate_discount function should apply a 10% discount when the order reaches $100."
Original PR code:
def calculate_discount(total_amount):
if total_amount > 100:
return total_amount * 0.10
return 0
- Read your context and instructions
- Compiled the sofware engineering rulebook
Engineering Rulebook — Essentials
PASS 1 — INSPECT
Finding: Incorrect threshold comparison
- Severity: Medium — Incorrect business logic
- Location:
discount_service.py,calculate_discount() - Root cause: The strict
>comparison excludes the exact $100 threshold. - Impact: An order totaling exactly $100 receives no discount, contradicting the specified business rule.
Boundary analysis:
| Order total | Expected discount | Current result |
|---|---|---|
| $99.99 | $0.00 | $0.00 |
| $100.00 | $10.00 | $0.00 ❌ |
| $100.01 | $10.001 | $10.001 |
Engineering principles applied: Boundary correctness, minimal changes, and regression evidence.
Recommendation: Change the comparison to include the threshold. No broader refactoring is justified.
PASS 2 — FIX
Proposed minimal patch:
def calculate_discount(total_amount):
- if total_amount > 100:
+ if total_amount >= 100:
return total_amount * 0.10
return 0
Change scope:
- One operator modified.
- No new dependencies.
- No unnecessary abstractions.
- No changes to unrelated business logic.
PASS 3 — VERIFY
Focused regression test:
import pytest
@pytest.mark.parametrize(
"amount, expected",
[
(99.99, 0),
(100, 10),
(100.01, 10.001),
],
)
def test_discount_boundary(amount, expected):
assert calculate_discount(amount) == pytest.approx(expected)
Expected verification sequence:
| Check | Before fix | After fix |
|---|---|---|
| Below threshold | PASS | PASS |
| Exactly at threshold | FAIL | PASS |
| Above threshold | PASS | PASS |
These are expected outcomes for the demonstration, not results from an executed test run.
Engineering Summary
- Patch: One comparison operator
- Regression coverage: Three boundary cases
- Scope expansion: None
- Verification status: Tests proposed; execution pending
The agent should run the regression tests and report the actual outcomes before declaring the repair verified.
The Difference
Generic AI approach:
"Change > to >=. Fixed!"
Engineering Rulebook approach:
Establish the defect, identify the relevant engineering principle, make the smallest justified correction, test the boundary, and report observable evidence.
One defect. One minimal patch. A repeatable engineering discipline.
Essentials handles focused repairs. Extended brings the full multi-pass engineering workflow to cross-file changes, larger features, and deeper verification.
Connects securely to your tools. The creator never sees your data.
About this skill
Software Engineering Rulebook — Essentials
Build more reliable code with a disciplined, literature-grounded approach to AI-assisted development.
Extended and Pro versions are also available.
AI coding agents can produce convincing solutions that introduce subtle bugs, unnecessary complexity, or unrelated changes. Software Engineering Rulebook — Essentials helps prevent these problems by applying established software engineering principles to code review and targeted defect repair.
Built around a structured Inspect → Fix → Verify workflow, Essentials guides your coding agent toward small, intentional changes supported by reproducible evidence and focused regression testing.
What It Does
- Literature-grounded code review: Identifies up to three significant issues, prioritizes them by risk, and connects recommendations to established engineering principles and rule references.
- Targeted defect repair: Diagnoses and fixes one localized issue through a disciplined three-stage workflow.
- Minimal, controlled patches: Limits modifications to the smallest necessary scope, reducing accidental regressions and unnecessary refactoring.
- Evidence-based verification: Establishes the expected failure before applying a correction, then verifies the result through focused regression checks.
- Essential security safeguards: Checks critical trust boundaries, including exposed secrets, unsafe input handling, and sensitive information disclosure.
How It Works
1. Inspect — Understand before changing
Analyze the relevant code, identify the underlying problem, and establish observable evidence of the defect. For repair tasks, create or identify a reproducible failing test whenever feasible.
2. Fix — Make the smallest effective change
Apply a focused correction that respects existing architecture, coding conventions, and established software engineering practices. Avoid unrelated modifications, speculative improvements, and unnecessary abstractions.
3. Verify — Demonstrate the result
Run applicable regression checks, compare observed behavior against expectations, and report precisely what was verified. Reference relevant engineering rule IDs and explicitly identify any remaining uncertainty.
Supported Languages & Frameworks
Designed for use with AI coding agents across languages and frameworks, including:
- JavaScript and TypeScript
- Python
- Go
- Other languages with independently testable logic
The workflow is framework-agnostic and supported by a curated reference library of literature-grounded software engineering rules.
Why Use It Instead of Generic AI Prompts?
A generic instruction such as "fix this bug" leaves your coding agent free to reinterpret the problem, modify unrelated code, or declare success without sufficient evidence.
Essentials introduces a repeatable engineering discipline that emphasizes:
- Evidence over assumptions
- Minimal changes over unnecessary rewrites
- Verification over unsupported claims
- Established engineering principles over improvised solutions
The result is a more controlled, reviewable, and reliable AI-assisted coding workflow.
Ideal Use Cases
- Reviewing AI-generated code or pull requests for significant logic defects.
- Correcting isolated bugs, edge cases, and missing guard conditions.
- Adding regression tests for previously reported failures.
- Reducing unnecessarily large patches to focused corrections.
- Improving the consistency and reliability of everyday AI-assisted development.
Scope & Limitations
Essentials is optimized for localized defects and focused code reviews.
It does not include large-scale refactoring, database schema migrations, complex concurrency engineering, advanced performance optimization, or CI/CD orchestration.
For broader engineering tasks, Software Engineering Rulebook — Extended adds a comprehensive multi-pass development workflow, dedicated frontend and backend engineering guidance, cross-file changes, and more extensive verification.
For production-level engineering, Software Engineering Rulebook — Pro extends the system with advanced security, computation optimization, Docker deployment, and repository-level governance.
Start with the fundamentals. Make every change intentional. Verify what actually works.
Handles local changes to one module only. No support for multi-file refactors, public schema migrations, or complex distributed systems. Requires host-provided tools for actual test execution.
How to use
- 1
Connect your agent to the Agensi MCP
Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup
- 2
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 3
Ask for what you need
Your agent loads this skill and uses it on your task. It always asks before any payment.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill goes through the trust checks below.
Security scanned
Checked by our 8-point scan before listing
Fresh listing
Recently published to Agensi
Money-back guarantee
Doesn't do what its page promises? You get your money back.
Trust & safety
Security scan
Security scan passed today
Requires
Permissions
Allowed Hosts