Works with the AI tools you already use
Claude Code
Cursor
Codex CLI
GitHub Copilot
Gemini CLI+20 moreClaude CodeCursorCodex CLIGitHub CopilotGemini CLI
OpenClaw
Manus
Grok+17 agentsArchitectureAdvisor: Pattern Recommender
Your code feels wrong but you can't name why? This skill scans your files, names the pattern that fixes each problem, and points you to the tool that implements it.
Free
ArchitectureAdvisor: Pattern Recommender
Example session with this skill installed
My main.py is getting huge and the agent keeps breaking things. Is there a better way to structure this?
- Read your context and instructions
- Compiled the architectureadvisor
- Generated the UI component
I scanned your project and identified two critical architectural signals that are likely causing the agent issues.
Architectural Advisory
SIGNAL: Single file over 800 lines with mixed responsibilities (models, handlers, and utils).
PATTERN: Module Split.
SHAPE:
Move logic into a structured directory tree:
config.py(Environment/Settings)models.py(Schemas/Types)services/(Business logic)handlers/(Entry points/Routes)
Themain.pyshould only serve as the entry point that wires these modules together.
WHY IT MATTERS: AI agents lose context window efficiency around 1,500 lines. When files are too long, the agent will silently drop functions or hallucinate existing logic during edits.
SIGNAL: Direct os.getenv() calls scattered across database.py and auth.py.
PATTERN: Config Broker.
SHAPE:
# config.py
class Settings:
DATABASE_URL: str = os.getenv("DATABASE_URL")
API_KEY: str = os.getenv("API_KEY")
settings = Settings()
Every other module should import from this single source of truth.
WHY IT MATTERS: Secret rotation becomes impossible when keys are hardcoded in multiple leaf modules. It also prevents easy mocking during testing.
Next steps
- Create a
services/directory to move business logic out ofmain.py. - Consolidate all
os.getenvcalls into a newconfig.pyfile. - Ask for a specific implementation once the new file structure is created.
architectureadvisor-pattern-recommender.tsx
TSX · React component
Example file from a real run - the skill writes it into your workspace.
Connects securely to your tools. The creator never sees your data.
About this skill
Your codebase feels messy but you can't put a finger on why. This skill analyzes your files to identify structural weaknesses and maps them to industry-standard architectural patterns.
Unlike prompt-only advisors, this skill is hard-coded to recognise eight specific, high-impact signals — and it refuses to implement anything, so you stay in control.
Verified in a real run
SIGNAL: Single file over 800 lines with mixed responsibilities PATTERN: Module split — one file per responsibility SHAPE: Split into config.py, db.py, models/, services/, handlers/, utils/. Each file holds one category. The entry file only wires them. WHY IT MATTERS: the AI agent loses track of the full file around 1500 lines. Functions get dropped, helpers get re-invented, the same block gets rewritten twice in one session.
SIGNAL: Direct os.getenv() calls in multiple locations
PATTERN: Config broker — one module reads the environment
SHAPE: Create config.py with a single class that reads every variable.
Every other file imports from config import config.
WHY IT MATTERS: a leaked token is found in one place but referenced in ten.
SIGNAL: Bare requests.post() to an external API with no timeout PATTERN: Resilience wrapper — timeout + fallback SHAPE: Wrap the call in try/except with an explicit timeout. Add a second endpoint. Route to it on verified failure. WHY IT MATTERS: a single provider outage takes the app down.
What you get
- SKILL.md — the agent instructions.
- SKILL.zip — the packaged skill for install.
- 8 architectural signals the skill recognises.
- Structured output: SIGNAL / PATTERN / SHAPE / WHY IT MATTERS.
- Zero implementation — pure advisory.
What it does
- Detects anti-patterns: manual import rows, raw dict access at boundaries, scattered environment variables.
- Names the solution: the exact pattern name that bridges "code smell" and "proven fix".
- Shows the shape of the fix: short code fragments or structural descriptions.
- Explains risk: what breaks in production if the pattern is skipped.
- Refuses to implement: you stay in control.
Why this beats prompting it yourself
A general-purpose LLM rewrites the whole file or hallucinates abstractions. This skill is hard-coded to recognise eight specific signals and returns a structured advisory — nothing more, nothing less.
Continue with paid skills
If this advisor finds a pattern you want implemented, these paid skills do the work:
- APIFactory — router factory, dynamic route auto-registration.
- CommitComposer — Conventional Commits enforcement with a runnable checker and privacy guard.
Known limitations
- Does not scan projects under 300 lines.
- Does not perform the refactoring — advisory only.
- Not for use during active production incidents.
- Reports a maximum of three signals per response.
How to install
Works the same in every agent - Claude, Cursor, Codex, Copilot and 20+ more.
- 1
Download the ZIP
Free skills download straight away. Paid skills unlock right after purchase.
- 2
Unzip into your skills folder
Every agent reads skills from one folder on your machine. Drop the unzipped folder in there.
- 3
Ask your agent to use it
Restart the agent if it was already running. It picks the skill up automatically - no config needed.
Skills folder by agent
Click the path to copy it. Create the folder if it does not exist yet.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill passes our trust checks below.
Security scanned
Passed our 8-point scan before listing
Fresh listing
Recently published to Agensi
Free forever
No account required to browse
Trust & safety
Security scanned
Verified clean today
- Free to download with an account