LetsVibe
Local AI
Graded 100% locally on what people experience: value, creativity and the user journey
| Rank | Repository & Owner | Score & Grade | Archetype | Factor Balance | Merit Badges | Action |
|---|---|---|---|---|---|---|
| Loading rankings... | ||||||
| Evaluation Factor | Weight | Raw Score | Contribution | What we found |
|---|
Show your grade in your README or on your website. The badge always shows your latest grade and links to this report.
Code-health findings from an automatic scan, each with the file and line to check. They are notes for the developer and never count toward the score.
Not required to be a great vibe-coded project, but these are the steps that would take it further.
Pick an issue below and copy the prompt into your AI coding assistant to verify whether it is real and plan a fix in planning-only mode.
Ecosystem creators ranked by their highest-graded masterpiece or comprehensive portfolio average across all evaluated repositories.
| Rank | Creator | Peak Score & Grade | Primary Languages | Graded Repositories | Top Project |
|---|---|---|---|---|---|
| Loading creator rankings... | |||||
Preserves the permanent legacy of repositories that achieved podium standing across each evaluation dimension, retaining full historical merit.
Reports held before publishing: owner explanations of flagged lines (shown as plain text and never run; unanswered requests are deleted automatically after 7 days), and graded projects the vibe-coded check could not confirm (they wait for you, nothing is asked of the owner).
Loading...
Owner error reports the automatic check did not confirm. Applying a correction updates the live database and every backup.
Loading...
Control repository evaluation queue, pause/resume processing, manually trigger audits, manage projects in the database, and inspect local Ollama diagnostics.
No tasks queued.
Re-queue every repository stored in the database to re-evaluate them with your updated prompts. They will process sequentially through the background worker queue.
Re-runs only the build-file scan and the quick 7B classify call for every project, updating its Category, System and device profile filters. Scores are not changed. Takes roughly 5–15 seconds per project through the queue.
Permanently wipe all projects, evaluation reports, rankings, and queue tasks. Use this during testing iterations or right before going live.
Manage and delete projects, associated report histories, and badges from the database.
Loading developer hierarchy...
User appeals against non-vibe disqualifications. Fair FIFO order (oldest to newest). Violent or profane messages are automatically red-flagged and relegated to the bottom of the queue.
Loading appeals...
Authenticating with GitHub. This page will refresh when you're signed in.
Connect your GitHub account to submit repositories and claim perks
Welcome to the early access preview of LetsVibe! By auditing projects during this Community Beta, you are helping test the local AI grading engine, which grades what people experience when they use a project. All early beta contributors receive:
Sign in to submit your own repositories:
New sign-ups are paused during a private preview. Existing accounts can still sign in.
read:user only: your public username, ID and avatar, used to sign you in and
confirm which repositories are yours. LetsVibe never asks for or stores your email address.
Privacy policy
Enter the URL of a public GitHub repository owned by your account to have it evaluated across all 7 dimensions. To rank work from another GitHub account, sign in with that account. You can remove your projects at any time from My Data.
.zip, .exe, .tar, .bin, .pt) are automatically excluded.
Audits run on a single local GPU, so they take a few minutes. If you'd like to help keep LetsVibe free, you can support it on Ko-fi.
Choose the statement that is wrong and explain why, pointing to the relevant files or functions. It is checked against your code once; if the check agrees, the review is corrected everywhere. Otherwise it goes to manual review. Track it on My Projects.
Explain how this project was vibe-coded: which AI wrote it and how you prompted it (e.g. Claude, ChatGPT, Cursor, Copilot; your prompts or workflow). Appeals are reviewed in the order they arrive. The decision will appear on your My Data page; no email address is needed.