Works with the AI tools you already use
Claude Code
Cursor
Codex CLI
Muse
OpenClaw+21 moreClaude CodeCursorCodex CLIMuseOpenClaw
Manus
Grok
Grok Bot+18 agentsVibeCoderPlaybook: Project Stage Navigator
Architectural roadmap tool that diagnoses project bottlenecks and routes vibe-coders to the right refactoring step.
$19
VibeCoderPlaybook: Project Stage Navigator
Example session with this skill installed
My main bot script is 1,600 lines long. The agent is starting to delete old functions when I ask for new ones. I have 350 active users on SQLite. What should I do?
- Read your context and instructions
- Compiled the vibecoderplaybook
I have analyzed your project signals. You are currently hitting a specific architectural wall that is degrading your AI agent's performance.
Diagnosis: Stage 2 (Monolithic Bottleneck)
| Signal | Value | Status |
|---|---|---|
| Primary File Size | 1,600 lines | Critical (Threshold: 1,500) |
| Agent Behavior | Deleting functions | Context Loss |
| User Load | 350 Active | Stable for current DB |
| Infrastructure | SQLite | Appropriate for Stage 2 |
What is happening
Your project has entered Stage 2. The AI agent is not becoming "stupid," it is simply running out of context window. At 1,600 lines, the agent cannot hold the entire file in its working memory while processing your new requests. It is prioritizing your new instructions by "forgetting" or overwriting existing code to stay within its token limit.
What breaks if you skip this
Silent Deletions: The agent will delete helper functions you aren't currently looking at, leading to production bugs.
Instruction Drift: The agent will start ignoring style guidelines or security constraints established at the top of the file.
High Latency: Your development velocity will drop as you spend more time fixing what the agent broke than building new features.
Recommended Next Action
Split the project structure using FileTreeArchitect.
You must break the 1,600-line monolith into logical modules (e.g., routes/, services/, utils/). This allows the agent to focus on one small, high-context file at a time.
What comes after
Once the file is split, the agent's performance will stabilize. You are not yet ready for Stage 3 (Database Migration). SQLite will handle your 350 users comfortably until you reach ~1,000 concurrent writers. Focus entirely on the file structure now.
Next steps
- Install or invoke
FileTreeArchitectto map your new folder structure. - Move your utility functions into a separate
utils.pyorutils.jsfile first to reduce the main file size. - Verify the bot still runs after the first split before moving onto API routes.
Connects securely to your tools. The creator never sees your data.
What you get
About this skill
Vibe-coding often leads to a wall where the AI agent starts hallucinating, deleting functions, or losing context. This happens because the project has outgrown its current structure, but identifying the specific architectural bottleneck is difficult when you are moving fast.
What it does
- Diagnoses project maturity by analyzing file sizes, user counts, error patterns, and deployment methods to identify the current development stage.
- Identifies context bottlenecks specifically for AI-driven development, flagging when a monolith file is causing the agent to forget code.
- Routes to specialized tools by recommending the exact next skill needed, such as
FileTreeArchitectfor splitting code orDBMigratorfor scaling storage. - Prevents premature optimization by refusing to recommend complex refactors or migrations for projects that are not yet at the required scale.
How it works
- Signal Collection: You provide at least three data points regarding your lines of code, active user count, and current error types.
- Stage Mapping: The skill maps these signals against seven distinct project stages, from "Stable MVP" to "Monolithic Bottleneck" or "Database Limit."
- Action Routing: It delivers a single, prioritized recommendation for the next architectural step.
- Impact Assessment: It explains exactly what will break if you ignore the advice and what the project will look like after the fix.
Frameworks & tools
This skill is framework-agnostic and works with any stack including Node.js, Python, React, and databases like SQLite or PostgreSQL.
Why this beats prompting it yourself
Generic LLM prompts often suggest "best practices" that lead to over-engineering. This skill enforces a strict hierarchy of needs, ensuring you don't migrate your database when the real issue is actually a 2000-line main file killing your agent's context.
Use cases
- Determining if it is time to split a large single-file application.
- Deciding when to move from SQLite to a managed database like Postgres.
- Hardening a project after gaining the first paying customers.
- Troubleshooting why an AI agent has suddenly started repeating itself or deleting code.
Known limitations
Does not provide implementation code for features. Requires at least three specific project signals to function. Not for use during active server outages or incidents.
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
30-day refund
Not a fit? Get your money back
Trust & safety
Security scanned
Verified clean 1 day ago
- Passed all security checks, Safe to install