Monomind vs plain Claude Code subagents: what the extension adds
Claude Code already has subagents, custom agents, skills, slash commands, hooks and MCP support. Monomind adds what it lacks: memory that persists across sessions in local SQLite, a codebase knowledge graph for impact analysis, document retrieval injected into prompts, 88 ready-made agents and 40+ workflows, background agent orgs that keep running after you close the session, and a prompt-injection gate. Small projects often don't need it.
What Monomind adds
- Cross-session memory stored locally with local embeddings (no cloud vector database).
- Monograph: callers, callees and blast radius of a symbol before an edit.
- Second Brain: your Markdown, PDF and DOCX files searched and injected into relevant prompts.
- Org Runtime: policy-gated role hierarchies that run as a daemon with a live dashboard, budgets and approvals.
- MonoFence: local prompt-injection and jailbreak detection as a pre-bash / pre-write gate.
- One monomind init configures Claude Code, OpenCode, Antigravity, Kimi Code and Codex the same way.
What it costs
- More hooks, MCP tools and context in every session.
- Node.js 22.12+ and native dependencies (better-sqlite3, tree-sitter).
- A larger surface to learn.
Frequently asked questions
Does Monomind replace Claude Code?
No. Monomind runs on top of Claude Code (and other assistants). Parallel work is still done by the assistant's own subagents; Monomind coordinates them and adds memory, the code graph and workflows.
Is Monomind free?
Yes. It is open source under Apache 2.0 and installed with npm install -g monomind.