← Monoes

01 · TypeScript

Monomind

Local memory, a code graph and agent teams for your AI coding assistant.

5Coding Assistants
14Org Runtimes
198MCP Tools

Interactive Demo

Use casesSee an org build features, ship, and write the blog

Autonomous Org · dev-team

simulated
$
Activity log…
● idle

Monograph · Knowledge Graph

★ API★ DB★ UIgod nodes
Activity log…

AST-parsed dependency graph of any codebase. Louvain clustering groups files into logical communities. God nodes (API, DB, UI) are highly-connected hubs. Changing them has the widest blast radius.

Good for: Understanding blast radius before edits, finding dead code, routing agents to the right files automatically.

● idle

Key Features

Everything you need. Nothing you don't.

6 capabilities

01
🏢

Autonomous Orgs

Define roles, reporting lines and per-role tool, file and budget policy in one JSON file, then run it as a background daemon with human approval gates, cross-run memory and a live dashboard. more than a dozen runtimes, including the Claude Agent SDK, Codex, OpenCode, Kimi Code and Copilot.

02
🧠

Local Memory

Local SQLite with on-device embeddings (gte-modernbert-base) and an HNSW index that switches on automatically past 100,000 entries. No cloud vector database, no API key. Context survives across sessions, agents and orgs.

03
🗺️

Codebase Knowledge Graph

Monograph parses your code with tree-sitter (TypeScript, Python, Go, Rust, Java, C/C++, C#, Ruby, Swift, PHP, Kotlin, Dart and more) into a SQLite graph, so your assistant sees callers, impact and blast radius before it edits.

04
📚

Second Brain

Drop in Markdown, PDF or DOCX and the relevant excerpts are injected into every prompt by meaning, not keywords. Indexing and search run locally, with a personal global brain shared across projects; the excerpts added to a prompt go to your AI provider with it.

05
⚡

Mastermind Workflows

40 /mastermind slash commands (plan, execute, review, debug, release, research, worktree and more), plus 83 agents and ready-made skills installed into your project. Add --tillend to repeat until nothing is left to do.

06
🛡️

Guardrails & Hooks

MonoFence blocks prompt injection before shell commands and file writes, and gates catch destructive commands and secrets. Backed by 28 hook subcommands and 9 background workers that refresh at session start.

CLI Reference

Scriptable. Pipeable. AI-ready.

Everything is also available from the terminal. The commands you'll use most, grouped by job.

⚡

Inside Claude Code and the other supported assistants, the same features are exposed through the MCP server (20 core tools by default, 198 in full mode) and /mastermind:* slash commands, so you rarely need to type these yourself.

Setup & health

Install into a project and check it.

monomind init
monomind init --target codex
monomind mcp start
monomind mcp verify
monomind doctor --fix

Autonomous orgs

Run agent organizations as a background daemon.

monomind org create blog --template content-team --goal "3 posts/week"
monomind org validate blog
monomind org run blog --task "weekly report"
monomind org status
monomind org questions blog
monomind org stop blog

Memory

Local SQLite memory with semantic search.

monomind memory store --key "auth" --value "JWT with refresh" --namespace patterns
monomind memory search -q "authentication patterns"
monomind memory stats

Second Brain

Index and search your own documents locally.

monomind doc ingest ./notes
monomind doc search -q "pricing notes"
monomind doc list

Code graph

Build and query the Monograph knowledge graph.

monomind monograph build
monomind monograph stats
monomind monograph watch

Picking, health & security

Pick the best-fit agent and skill, check your setup, and scan.

monomind pick -t "fix the login bug"
monomind doctor
monomind security scan

$ monomind --help

Get Started

Up and running in seconds.

$ ▋