Help us improve
Share bugs, ideas, or general feedback.
Share bugs, ideas, or general feedback.
Share bugs, ideas, or general feedback.
Metacognitive advisor that monitors Claude Code and provides feedback
npx claudepluginhub open-horizon-labs/bottle --plugin superegoTemporarily disable superego evaluation for this session.
Enable superego for this project/session.
Initialize superego for this project.
Manage superego prompts for this project. Available actions:
Remove superego from this project (inverse of init).
Metacognitive review of software development work. Use when writing code, reviewing PRs, or making architectural decisions.
Learning coach reviewing teaching approaches. Use when an AI is helping someone learn or understand concepts, to ensure teaching is hands-on and verifiable.
Collaborative review of written content. Use when drafting blog posts, documentation, essays, or other written material.
Matches all tools
Hooks run on every tool call, not just specific ones
Uses power tools
Uses Bash, Write, or Edit tools
Share bugs, ideas, or general feedback.
Own this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge.
Sign in to claimOwn this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge.
Sign in to claimBased on adoption, maintenance, documentation, and repository signals. Not a security audit or endorsement.
Metacognitive advisor that monitors Claude Code and provides feedback
A subconscious for Claude Code. A Letta agent watches your sessions, accumulates context, and whispers guidance back.
Automatic Claude Code -> Obsidian second brain: zero-config session capture, hybrid search, autonomous recall, daily/lessons/preferences.
Persistent memory across Claude Code sessions using Cognis
Open-source, local-first Claude Code plugin for token reduction, context compression, and cost optimization using hybrid RAG retrieval (BM25 + vector search), reranking, AST-aware chunking, and compact context packets.
Complete AI coding workflow system. Self-correcting memory + persistent FTS5-indexed research wikis + auto-research loop + multi-LLM council on a single SQLite store. 33 skills, 8 agents, 22 commands, 37 hook scripts across 24 events. Cross-agent via SkillKit.
Unified Hi-Fi Control - Control Roon playback and HQPlayer audio pipeline
Metacognitive advisor that monitors Claude Code and provides feedback
Metacognitive advisor that monitors Claude Code and provides feedback
Simple task tracking for LLM sessions - ownership-based workflow for multi-agent coordination
Simple task tracking for LLM sessions - ownership-based workflow for multi-agent coordination
Share bugs, ideas, or general feedback.
Open Horizon Labs - One entry point for all the tools
Bottle provides the complete Open Horizon Labs stack for Claude Code, OpenCode, and Codex.
brew tap open-horizon-labs/homebrew-tap && brew install bottle
bottle install stable
bottle integrate claude_code # or: codex, opencode
If you have multiple Claude Code directories (e.g., ~/.claude for personal use and ~/.claude-work for work), use the CLAUDE_CONFIG_DIR environment variable:
bottle integrate --list # shows all detected directories
CLAUDE_CONFIG_DIR=~/.claude-work bottle integrate claude_code # integrate specific directory
Then initialize in your project:
/bottle:init # Claude Code
$bottle init # Codex
bottle-init # OpenCode
This installs binaries, initializes .ba/, .wm/, .superego/, and creates AGENTS.md.
Bottle is part of the Open Horizon Labs ecosystem - tools that give you AI superpowers without surrendering your mind. The ecosystem has two delivery mechanisms serving different audiences:
Open Horizons (strategic layer - missions, guardrails, metis)
|
+-- Bottle (developer tools)
| CLI tools that augment AI coding assistants
| You use these TO BUILD software
|
+-- Memex (end user product)
Desktop app that IS the AI interface
Knowledge workers use this directly
Bottle tools augment Claude Code, OpenCode, and Codex with:
Memex implements the same concepts natively for end users:
The philosophy is shared: grounding prevents drift, recall surfaces relevant knowledge, oversight catches mistakes, and strategic context keeps you aligned with what matters.
We use bottle tools to build Memex. Memex will be the finished product for knowledge workers who want AI without the command line.
For detailed installation and usage per platform:
Simple, file-based task tracking. No server, no database.
Commands: /ba:init, /ba:quickstart, /ba:status
Automatically captures tacit knowledge from sessions and injects relevant context.
Setup: Installs plugin + binary, works automatically
Monitors Claude's work and provides feedback before finishing or making large changes.
Commands: /superego:init, /superego:review, /superego:prompt
Builds a searchable knowledge graph from Claude Code sessions, making insights from past sessions queryable.
Commands: /datasphere:init, /datasphere:setup
Connects Claude Code to Open Horizons for strategic context.
Setup: /oh-mcp:setup (requires OH account + API key)
Telegram bot for running Claude sessions remotely.
Setup: Server component, see miranda docs
bottle update
See platform-specific docs for manual update commands.
Want to pin specific tool versions or create a team-shared configuration? Create a bespoke bottle:
bottle create mystack --from stable
# Edit ~/.bottle/bottles/mystack/manifest.json
bottle install mystack