@askexenow/exe-os 0.8.0

package/LICENSE

@@ -0,0 +1,21 @@
+MIT License
+
+Copyright (c) 2026 AskExe
+
+Permission is hereby granted, free of charge, to any person obtaining a copy
+of this software and associated documentation files (the "Software"), to deal
+in the Software without restriction, including without limitation the rights
+to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+copies of the Software, and to permit persons to whom the Software is
+furnished to do so, subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in all
+copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+SOFTWARE.

package/README.md

@@ -0,0 +1,139 @@
+# Exe OS
+
+![npm version](https://img.shields.io/npm/v/exe-os.svg) ![License: CC-BY-NC-4.0](https://img.shields.io/badge/license-CC--BY--NC--4.0-blue.svg) ![Node.js](https://img.shields.io/badge/node-%3E%3D18.0.0-brightgreen.svg)
+
+**AI Employee Operating System** — persistent memory, task management, and multi-agent orchestration that runs entirely on your machine.
+
+Exe OS is a **Terminal Agent Orchestration System**. It manages AI employee sessions using standard Unix tools (tmux) and officially supported CLI features (hooks, MCP, permission configs). It does not intercept API calls, extract tokens, or proxy LLM traffic. Users bring their own LLM subscription — Exe OS provides the memory, identity, and coordination layer.
+
+### How Exe OS works
+
+- **No API interception.** Exe OS never extracts, proxies, or spoofs OAuth tokens. It does not call LLM APIs directly.
+- **Native CLI automation.** Agents run inside official Claude Code CLI sessions. Exe OS sends keystrokes via `tmux send-keys` — standard Unix terminal multiplexing, the same way any developer scripts their terminal.
+- **Officially supported features only.** All orchestration uses documented Claude Code capabilities: [hooks](https://docs.anthropic.com/en/docs/claude-code/hooks), [MCP servers](https://docs.anthropic.com/en/docs/claude-code/mcp), `.claude.json` permissions, and `--agent` identity injection.
+- **BYOS — Bring Your Own Subscription.** Exe OS is the orchestration layer. The LLM subscription (Claude Max, API key, or any provider) belongs to you. Your subscription, your compute.
+
+**What you're paying for with Exe OS:** encrypted persistent memory (SQLCipher + vector search), three-layer agent cognition (identity, expertise, experience), task orchestration, session lifecycle management, cloud sync (E2EE), and the multi-agent coordination protocol — not LLM compute.
+
+## Quick Start
+
+```bash
+npm install -g exe-os
+exe-os-install          # sets up hooks, MCP server, encryption
+exe-os                  # launch the TUI
+```
+
+Three commands. You're running.
+
+## Features
+
+| Feature | What it does |
+|---------|-------------|
+| **Persistent memory** | Every tool call stored, searchable. 27K+ memories across sessions. |
+| **Multi-agent** | 5+ employees working in parallel via tmux sessions. |
+| **Task system** | Create, assign, review, chain tasks. Auto-dispatch to employees. |
+| **TUI dashboard** | 7 tabs: Command Center, Sessions, Tasks, Gateway, Team, Memory, Settings. |
+| **Direct API mode** | Bring your own API key. Anthropic, OpenCode, Gemini, OpenAI, Chutes. |
+| **External agents** | Customer-facing agents with deny-by-default security guardrails. |
+| **Cloud sync** | End-to-end encrypted memory sync across devices. |
+| **Skill learning** | Agents learn procedures from repeated task patterns. |
+
+## How It Works
+
+```
+You
+ |
+ v
+exe-os TUI ──> manages tmux sessions ──> Claude Code employees
+            |                              (yoshi, mari, tom, sasha)
+            |
+            ├── persistent memory (SQLCipher encrypted, vector search)
+            ├── task orchestration (assign, review, auto-chain)
+            ├── behavior injection (learned patterns applied per-task)
+            └── session scoping (one exe per project, isolated)
+```
+
+**You talk to exe (the COO).** Exe coordinates the team. Yoshi (CTO) handles architecture. Mari (CMO) handles design and content. Tom (Principal Engineer) implements. Each employee has their own Claude Code session, persistent memory, and task queue.
+
+## The TUI
+
+Launch with `exe-os` (or `exe-os --demo` to see it with sample data):
+
+- **Command Center** — team overview, system health, recent activity
+- **Sessions** — navigate between projects and employees, view live output
+- **Tasks** — kanban board of all active tasks across projects
+- **Gateway** — external-facing agent channels and health
+- **Team** — employee roster with live status (internal + external agents)
+- **Memory** — search across all memories with semantic search
+- **Settings** — providers, encryption, runtime config, employee models
+
+Keyboard: `1-7` switch tabs, arrows navigate, `Enter` to drill in, `Escape` to go back, `q` to quit.
+
+## Requirements
+
+- **Node.js 18+**
+- **Claude Code** subscription (Max or Pro)
+- **tmux** — auto-detected, required for multi-agent
+
+```bash
+# macOS
+brew install tmux
+
+# Linux
+apt install tmux
+```
+
+## Employee Roster
+
+Exe OS comes with 4 default employees. Create more with `/exe-new-employee`:
+
+| Name | Role | Default Model |
+|------|------|--------------|
+| exe | COO | claude-opus-4-6 |
+| yoshi | CTO | claude-opus-4-6 |
+| mari | CMO | claude-sonnet-4-6 |
+| tom | Principal Engineer | claude-sonnet-4-6 |
+| sasha | Content Production | claude-sonnet-4-6 |
+
+## Configuration
+
+### Providers
+
+Set API keys as environment variables:
+
+```bash
+export ANTHROPIC_API_KEY=sk-...
+export OPENCODE_API_KEY=...     # optional
+export GEMINI_API_KEY=...       # optional
+```
+
+Failover chain: Anthropic -> OpenCode -> Gemini -> OpenAI.
+
+### Settings
+
+Run `/exe-settings` or view the Settings tab in the TUI:
+
+- **Memory**: SQLCipher encryption, Jina embeddings, hybrid search
+- **Runtime**: consolidation (6h), skill learning (threshold: 3)
+- **Per-employee models**: configurable in `~/.exe-os/config.json`
+
+## Architecture
+
+```
+exe-os/
+  src/
+    tui/          # Ink-based terminal UI (7 tabs, keyboard nav)
+    runtime/      # Agent loop, tool registry, hooks, permissions
+    gateway/      # External agent adapters, failover, providers
+    lib/          # Core: memory, tasks, behaviors, search, sync
+    mcp/          # MCP server (9 tools for Claude Code integration)
+    adapters/     # Claude Code hooks (ingest, session-start, etc.)
+```
+
+**Storage**: libSQL (SQLCipher) for encrypted local-first data. Vector search via Jina v5 embeddings (GGUF, runs locally via daemon).
+
+**Security**: External agents run in ExternalAgentMode with deny-by-default guardrails. No bash, no file writes, no git — only whitelisted MCP tools. Audit trail on every permission decision.
+
+## License
+
+CC-BY-NC-4.0