Skip to main content
Hopsule is the governance infrastructure for AI-assisted development. Remember, enforce, and orchestrate your decisions, architecture, and context so people and AI tools can stop re-explaining the stack on every thread and start shipping.

​
What Hopsule is

Hopsule is the persistent decision and memory layer for AI-assisted development. Architectural choices live as structured decisions; the why lives in memories. Capsules bundle the subset you want active. Brain is the interactive knowledge graph. Hopper is the in-app assistant that stays grounded in that context. Hopsule does not store your source code. It stores decisions, memories, and structural metadata your AI and IDE need, while the codebase stays in Git. Enforcement is advisory: Hopsule surfaces warnings and context when something conflicts with a team decision. It does not block merges, reject PRs, or auto-fix code for you. Humans decide; Hopsule informs. For how accepted decisions are stored immutably and how clients reflect them, see Decisions.

​
At a glance

  • Structured governance: decisions and memories in a durable layer, not lost chat history.
  • Portable Capsules: one context pack for Cursor, ChatGPT, Claude, Gemini, teammates, and automation.
  • Warnings where you work: IDE add-on diagnostics plus MCP so agents read the same rules.
  • Brain (knowledge graph): see how decisions, memories, tasks, and conflicts connect before you change course.
  • Plugs into your stack: Web app, IDE add-on, MCP server, CLI tool, and GitHub App (repository-level sync). See Platform overview.

​
How it works

1

Capture and review decisions

You record decisions and memories. They move through review: Draft, Pending, Accepted, Deprecated. AI can propose; your team approves when someone has permission to advance the lifecycle.
2

Bundle Capsules

You group related decisions and memories into Capsules, portable context packs you activate per project or workflow.
3

Use everywhere

You load Capsules and connect IDE, MCP, or CLI so assistants get persistent context without copy-pasting specs each time.

​
Why teams adopt it

Faster iteration

AI starts from accurate, up-to-date context; less back-and-forth in every conversation.

Higher consistency

Output can align with accepted decisions across tools and developers.

Lower token burn

Fewer repeated “here is our stack” preambles; reuse Capsule-backed context instead.

​
Product surfaces

Platform overview

Web app, IDE add-on, MCP, CLI, and GitHub App on one map.

Quickstart

Account, GitHub, Cursor, and first Capsule loop.

MCP server

Native tools and context for MCP-capable assistants.

CLI tool

Terminal workflows and sync.