- Home
- Skills
- Code Quality & Review
- No-BS Code Review (不废话代码审查)
Works with the AI Agents you already use
No-BS Code Review (不废话代码审查)
Pragmatic pre-merge code review that enforces redlines unconditionally, stops over-engineering, and respects project conventions. Strictly evaluates code without unsolicited modifications.
Free
No-BS Code Review (不废话代码审查)
Example session with this skill installed
Review this pull request for pre-merge quality gates. Focus on code complexity, over-engineering, and security.
- Read your context and instructions
- Compiled the no-bs code review
I have reviewed the registration service. While the logic is correct, the validation layer introduces unnecessary abstraction.
Verdict: Mergeable with Minor Fixes
The core logic is sound, but a mandatory fix is required for nesting depth in the password hashing block.
🔴 Must Fix (1 item)
-
services/user_service.py:42— Deeply nested conditional logic.
Reason: The if / try block is nested 4 levels deep, violating the maximum depth of 3. This reduces readability and makes error tracing difficult.
- Recommendation:
# Before
if user_data:
if validate_email(user_data.email):
try:
# hashing logic
except Exception:
# error
# After
if not user_data or not validate_email(user_data.email):
return False
try:
# hashing logic
except Exception:
# error
🟡 Consider (1 item)
-
wrappers/validator.py:10— Generic validation wrapper.
Note: This adds a polymorphic class hierarchy for a single concrete implementation. If this logic isn't reused across 3+ sites, a simple function is preferred over a generic wrapper.
💡 Minor Hints (1 item)
services/user_service.py:15— The variableudis a bit vague; consideruser_record.
Unchecked Scope
Database unique constraint conflicts were not verified against the schema.
Next steps
- Flatten the nested logic in the hashing block.
- Decide if the validation wrapper is truly necessary for future modules.
Connects securely to your tools. The creator never sees your data.
About this skill
No-BS Code Review
Pragmatic, strict pre-merge engineering reviews that eliminate fluff, block real defects, and put an end to over-engineering.
💡 Why No-BS Code Review?
Most automated AI code reviewers do more harm than good:
- Pedantic Nitpicking: Flooding PRs with trivial variable naming suggestions and syntax sugar debates that drown out fatal bugs.
- Dogmatic Metrics: Enforcing arbitrary line counts that force developers to chop cohesive methods into fragmented classes with sprawling state.
- Over-Engineering Bias: Suggesting Strategy patterns, Builder patterns, and abstract factories where a flat
switchor simpleif/elseis vastly superior. - Unsolicited Disruptions: Modifying code without permission, breaking existing codebase conventions and team standards.
No-BS Code Review is built from the battle-tested practices of senior software engineers, anchored by a single arbitration priority chain:
Correctness > Readability > Debuggability > Project Consistency > Maintainability > Simplicity > Performance > Elegance
🔍 Four-Pass Review Engine
-
Pass 1: 🔴 Redlines (Mandatory Merge Blockers)
- Deep nesting > 3 levels (
if/for/try) - Nested closures/lambdas or nested ternary operators
- Unjustified new third-party dependencies introduced into legacy codebases
- Concurrency race conditions, swallowed exception traces, and logic flaws
- Critical security vulnerabilities (SQLi, XSS, SSRF, broken auth, hardcoded secrets)
- Deep nesting > 3 levels (
-
Pass 2: 🟡 Yellow Lines (Reflection Triggers, NOT Mandatory Splits)
- Method length (~60 lines), new file length (~400 lines), chained calls (~5 stages)
- The 3-Sentence Self-Check: Can the developer explain the execution flow in 3 plain sentences? If yes, keep it as is.
- No artificial splitting: A flat 100-line
switchis far easier to debug than a 30-line triple-nested loop.
-
Pass 3: 🛡️ Anti-Over-Engineering
- Favor
switchover polymorphic hierarchies - Favor
if/elseover Strategy patterns; favorMap<String, Handler>over Registries - Direct constructors when parameters ≤ 4 rather than handwriting custom Builders
- Reject unnecessary domain layers (CQRS, Event Sourcing) for simple CRUD services
- Favor
-
Pass 4: 🛠️ Language & Domain Deep Dives
- Java: Enforce Stream single-link complexity rules, proper
Optionalsemantics, and context-rich logging. - TypeScript & Vue: Enforce type complexity boundaries ("if the type is harder to read than the code, simplify it with
as"), Composition API guidelines, and strict cleanup of memory leaks. - Backend & Persistence: Guard against N+1 query traps, unpaginated bulk queries, and network calls held inside database transactions.
- Java: Enforce Stream single-link complexity rules, proper
📋 Structured Output Format
Every review returns a clear, tiered, actionable summary:
- Verdict: Mergeable | Mergeable with Minor Fixes | Needs Rework
- 🔴 Must Fix: Exact file/line with engineering justification and minimal diff.
- 🟡 Consider: Suggestions that the author may keep if sound justification exists.
- 💡 Minor Hints: Maximum 2 non-blocking minor notes.
- Unchecked Scope: Transparent statement of what was not verified.
⚠️ Boundary Guarantee: This skill evaluates code only and will never modify your codebase unless explicitly commanded to apply fixes.
Does not execute code or run test suites. Will not modify files without an explicit command. Cannot verify live database index performance.
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
Free forever
No account required to browse
Trust & safety
Security scan
Security scan passed 10 days ago
- Free with an Agensi account
Requires