graybyrd13/claude-basecamp — explained in plain English
Analysis updated 2026-05-18
Set up a standing check that keeps a repository's test suite passing automatically.
Track monthly AI spending per repository so autonomous fixes never run up a surprise bill.
Resume a Claude Code session that crashed mid-task without losing its context.
Search every past Claude Code session on your machine to find a specific past conversation or command.
| graybyrd13/claude-basecamp | abhagsain/bayut-maps-ratings | abhishek-kumar09/who-is-imposter | |
|---|---|---|---|
| Stars | 3 | 3 | 3 |
| Language | JavaScript | JavaScript | JavaScript |
| Last pushed | — | — | 2021-04-12 |
| Maintenance | — | — | Dormant |
| Setup difficulty | easy | moderate | easy |
| Complexity | 3/5 | 2/5 | 3/5 |
| Audience | vibe coder | general | general |
Figures from each repo's GitHub metadata at analysis time.
Requires Node.js 18+ and an existing Claude Code setup, no database or extra config needed.
Claude Basecamp is a local dashboard that gives every one of your code repositories a persistent manager built around Claude Code, the AI coding assistant. You run it with a single command, npx claude-basecamp, and it opens at a local web address on your own machine, with no install, no database, and no configuration needed. It works on macOS, Linux, and Windows, and automatically finds the projects Claude Code already knows about. The main idea is what the README calls a reconciliation loop, borrowed from how systems like Kubernetes keep infrastructure in a desired state. You declare things that should always stay true about a repository, such as tests always passing or the changelog covering every release, and a background process keeps checking reality against that goal. When something drifts, it dispatches Claude Code to fix the problem, verifies the fix actually worked, and only asks you to step in for decisions a human truly needs to make. Spending is controlled through monthly budgets tracked per repository and per check, using the real dollar cost reported by the Claude command line tool rather than an estimate, so autonomous runs pause with a notice instead of running up a surprise bill. A background service option keeps checks and scheduled routines running even without a terminal open, catching up automatically after your computer sleeps. Checks can be written down in a versioned manifest file inside a repository, so cloning that repository lets a teammate see and choose to adopt the same standing checks. New automated fixes are made in an isolated git working copy on their own branch, so they never collide with work you are actively editing, and you review each result as a diff before applying or discarding it. The tool also learns from moments where you corrected or interrupted Claude, turning them into rules that block the same mistake happening again across every session on your machine. It keeps a searchable record of every past Claude session, can resume a session that crashed partway through unfinished work, and lets you talk to one persistent manager agent that has visibility across all your repositories and can adjust its own settings on request. It is released under the Unlicense, placing the code in the public domain.
A local dashboard that gives every code repository a persistent Claude Code manager that watches for drift, fixes it, and asks for help only when needed.
Mainly JavaScript. The stack also includes JavaScript, Node.js, Claude Code.
The code is placed in the public domain, you can use it for any purpose with no restrictions.
Setup difficulty is rated easy, with roughly 5min to a first successful run.
Mainly vibe coder.
This repo across BitVibe Labs
Don't trust strangers blindly. Verify against the repo.