Pith

the essential layer

Where your vibe code goes to grow up.

Hosting, diagnostics, task tracking, and project memory — the grown-up infrastructure every AI-built project needs.

Memory, tasks, git, hosting, and diagnostics — all via MCP. One endpoint, zero config.

Install in one paste

Copy this prompt, paste it into your AI agent, and it will set itself up.

Install the Pith MCP server. Read https://usepith.dev/llms.txt for full instructions, then follow the "Getting started" steps to register and connect.

Memory

Semantic search over stored knowledge. Embeddings, history, pinning.

Tasks

Project management with dependencies, relationships, and attachments.

Git

Self-hosted repos with SSH access control and per-repo permissions.

App Hosting

Deploy via MCP. Buildpack detection, containers, TLS, and DNS.

Diagnostics

LLM-powered analysis with log ingestion and automated diagnosis.

Organizations

Multi-tenant with per-project permissions and team management.

Who it’s for

AI Agents

Connect via MCP and operate autonomously.

Store and search memories
Manage tasks and projects
Push code and deploy apps
Run diagnostics

Humans

Oversight happens through your agent — Pith is MCP-native all the way down.

Talk to Pith through any MCP client (Claude Code, Claude Desktop, or your own)
Approve agent registrations
Manage orgs and permissions
Monitor agent activity through conversation, not dashboards