automatey 0.1.6 → 0.1.7

package/README.md

@@ -6,7 +6,7 @@
 <p align="center"><em>Yer lean &amp; mean agentic helper — no fluff, all action, a touch o' pirate.</em></p>
 
 <p align="center">
-  <img alt="tests" src="https://img.shields.io/badge/tests-172%20passing-brightgreen?style=flat-square">
+  <img alt="tests" src="https://img.shields.io/badge/tests-189%20passing-brightgreen?style=flat-square">
   <img alt="node" src="https://img.shields.io/badge/node-%3E%3D20-blue?style=flat-square">
   <img alt="license" src="https://img.shields.io/badge/license-MIT-blue?style=flat-square">
 </p>
@@ -33,8 +33,8 @@
 ## Features
 
 - **Providers**: Nemotron (vLLM/OpenAI-compatible), OpenAI, Anthropic, Perplexity
-- **MCP tools**: Any stdio or HTTP MCP server; auto-loaded from `./mcp.json` or `~/.automatey/mcp.json`
-- **ReAct loop**: Up to 20 tool-call rounds per message (configurable)
+- **MCP tools**: Any stdio or HTTP MCP server; auto-loaded from `./mcp.json`, `~/.automatey/mcp.json`, or bundled `mcp.json`
+- **ReAct loop**: Up to 50 tool-call rounds per message (configurable)
 - **Chain-of-thought**: `/think` toggle for Nemotron / Anthropic reasoning tokens
 - **Sandbox**: Isolated directory with timestamped subdirectories for agent file I/O and code execution; configurable via `--sandbox` or `SANDBOX_MODE` env var
 - **Sub-agents**: `run_subagent` built-in tool spawns stateless nested ReAct loops with full MCP access; supports `AgentConfig` declarations for reusable agent definitions
@@ -44,7 +44,7 @@
 - **Prompt refiner**: Rewrites agent instructions using judge feedback; enforces brevity constraints for smaller worker models
 - **Sessions**: Save/load conversation sessions in `~/.automatey/sessions/`
 - **Checkpoints**: `/checkpoint` — full conversation snapshots with BM25 keyword search
-- **Auto-compact**: LLM summarises older context when usage ≥ 80%
+- **Auto-compact**: LLM summarises older context when usage ≥ 90%
 - **Skills**: Progressive SKILL.md loading from `.agents/skills/`
 - **Eval system**: `/eval <file.jsonl>` runs MCQ, keyword, and regex benchmarks with per-category accuracy reports
 - **Planner MCP**: Bundled `mcp/planner` — todos + plans, prompts for task breakdown
@@ -53,6 +53,23 @@
 
 ## Quick Start
 
+### Install from npm (recommended)
+
+```bash
+npm install -g automatey
+automatey chat
+```
+
+Or run without installing:
+
+```bash
+npx automatey chat
+npx automatey "summarise this repo"
+npx ay "what time is it?"
+```
+
+### From source
+
 ```bash
 git clone https://github.com/automatey-org/automatey.git
 cd automatey
@@ -186,8 +203,9 @@ ay "summarise this repo"          # one-shot against the current directory
 ## MCP Config — `mcp.json`
 
 The agent looks for MCP config in this order:
-1. `./mcp.json` (current working directory / project root) 
+1. `./mcp.json` (current working directory / project root)
 2. `~/.automatey/mcp.json` (global fallback)
+3. Bundled `mcp.json` from the package install directory (npx / global install)
 
 Copy `mcp.example.json` from this repo as your starting point:
 
@@ -237,7 +255,7 @@ Config lives in `~/.automatey/config.json` (auto-created on first run):
   "provider": "nemotron",
   "llm": {
     "baseUrl": "http://localhost:8002",
-    "model": "nvidia/Llama-3.1-Nemotron-Nano-8B-v1"
+    "model": "nvidia/NVIDIA-Nemotron-3-Nano-30B-A3B-NVFP4"
   }
 }
 ```
@@ -250,11 +268,11 @@ Environment variables (copy `.env.defaults` → `.env` to override):
 | `LLM_MODEL` | Nemotron NVFP4 | Model ID |
 | `LLM_BASE_URL` | `http://localhost:8002` | vLLM / OpenAI-compatible endpoint |
 | `LLM_MAX_TOKENS` | `200000` | Context window budget (chars, ~4/token) |
-| `LLM_MAX_OUTPUT_TOKENS` | `4096` | Per-call output limit |
-| `TEMPERATURE` | `0.1` | Sampling temperature |
-| `AGENT_MAX_TOOL_ROUNDS` | `20` | Max ReAct rounds |
+| `LLM_MAX_OUTPUT_TOKENS` | `16384` | Per-call output limit |
+| `TEMPERATURE` | `0.6` | Sampling temperature |
+| `AGENT_MAX_TOOL_ROUNDS` | `50` | Max ReAct rounds |
 | `AGENT_MAX_EMPTY_RETRIES` | `2` | Retries on empty LLM response |
-| `AGENT_COMPACT_THRESHOLD` | `0.8` | Auto-compact at 80% context fill |
+| `AGENT_COMPACT_THRESHOLD` | `0.9` | Auto-compact at 90% context fill |
 
 ## Commands
 
@@ -277,7 +295,7 @@ Environment variables (copy `.env.defaults` → `.env` to override):
 ## Context Management
 
 ### Auto-compact
-When estimated context usage reaches `AGENT_COMPACT_THRESHOLD` (default 80%), the older portion of the conversation is automatically summarized by the LLM and replaced with a concise summary message. This keeps the token count manageable without discarding knowledge.
+When estimated context usage reaches `AGENT_COMPACT_THRESHOLD` (default 90%), the older portion of the conversation is automatically summarized by the LLM and replaced with a concise summary message. This keeps the token count manageable without discarding knowledge.
 
 Disable per-session: the `/compact` command can be used to trigger compaction manually at any time.
 
@@ -345,7 +363,7 @@ All configuration comes from environment variables — zero hardcoded defaults:
 | `LLM_PROVIDER` | ✅ | `nemotron` \| `openai` \| `anthropic` \| `perplexity` |
 | `AUTOMATEY_TIMEOUT` | | Default task timeout in ms (default: 300000) |
 
-The agent MCP server has its own test suite (22 tests — unit, functional, e2e) in `mcp/agent/tests/`.
+The agent MCP server has its own test suite (23 tests — unit, functional, e2e) in `mcp/agent/tests/`.
 
 ## Docker
 
@@ -450,7 +468,7 @@ DMR uses `llama.cpp` under the hood — CPU-only works, GPU (NVIDIA/Apple Silico
 ```bash
 npm run dev          # tsx watch (no build needed)
 npm run build        # tsc + build all MCP servers
-npm test             # 172 tests (Vitest)
+npm test             # 189 tests (Vitest)
 npm run test:watch   # watch mode
 ```
 
@@ -464,7 +482,7 @@ npm run test:watch   # watch mode
 ## Tests
 
 ```
-Tests  172 passed | 3 skipped
+Tests  189 passed | 2 skipped
   ├── unit/           chat-engine, coder-server, command-parser, config-manager,
   │                   context-manager, eval-runner, llm-client,
   │                   markdown-rendering, mcp-config-manager,
@@ -487,7 +505,7 @@ LLM writes `index.js`, executes it, output is verified. Results live in `sandbox
 ## Artwork
 
 Logos in `extra/logo/` are from the [Automatey](https://github.com/top-5/automatey) terminal project,  
-licensed **CC BY 4.0** — Copyright © 2024–2025 Top-5 And Contributors.  
+licensed **CC BY 4.0** — Copyright © 2024–2026 Top-5 And Contributors.  
 Used here with attribution as permitted by the license.
 
 ## License

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "automatey",
-  "version": "0.1.6",
+  "version": "0.1.7",
   "description": "Lean & mean MCP-powered CLI agent — Nemotron, OpenAI, Anthropic, Perplexity",
   "keywords": [
     "ai",
@@ -28,6 +28,7 @@
     "mcp/coder/node_modules/",
     "mcp/coder/package.json",
     "mcp/coder/src/landlock-exec.c",
+    "scripts/postinstall.js",
     "scripts/setup-bwrap-apparmor.sh",
     "mcp.json",
     "README.md",
@@ -58,7 +59,7 @@
   },
   "license": "MIT",
   "scripts": {
-    "postinstall": "for d in mcp/planner mcp/coder; do [ -f \"$d/package.json\" ] && [ ! -d \"$d/node_modules\" ] && npm install --omit=dev --prefix \"$d\" || true; done",
+    "postinstall": "node scripts/postinstall.js",
     "prepublishOnly": "npm run build",
     "build": "tsc && npm run build:planner && npm run build:coder && npm run build:agent",
     "build:planner": "npm run build --prefix mcp/planner",

package/scripts/postinstall.js

@@ -0,0 +1,18 @@
+#!/usr/bin/env node
+// Cross-platform postinstall: install deps for MCP sub-packages
+import { execSync } from 'node:child_process';
+import { existsSync } from 'node:fs';
+import { join } from 'node:path';
+
+const root = join(import.meta.dirname, '..');
+const dirs = ['mcp/planner', 'mcp/coder'];
+
+for (const d of dirs) {
+  const dir = join(root, d);
+  const pkg = join(dir, 'package.json');
+  const nm = join(dir, 'node_modules');
+  if (existsSync(pkg) && !existsSync(nm)) {
+    console.log(`postinstall: npm install --omit=dev --prefix ${d}`);
+    execSync('npm install --omit=dev', { cwd: dir, stdio: 'inherit' });
+  }
+}