Stable Release — v0.3.6

NARE CLI
SEMANTIC
MEMORY

Neural Amortized Reasoning Engine. An O(log N) persistence layer for software engineering. Amortized reasoning through semantic memory indexing.

Traditional assistants re-calculate identical logic at every prompt. NARE implements a 5-tier routing architecture that prioritizes locally-indexed skills and cached reasoning before falling back to full synthesis.

Install Now Features Documentation

O(log N)

memory lookup

5-Tier

routing architecture

0.2s

cached inference

nare — session

◆ NARE reasoning agent for software engineering

NareCLI ~/projects/nare

Autopilot mode · type /help for commands, Tab to cycle modes

> find and fix the authentication bug in auth.py and login.py

●Intent: edit

●Plan (moderate)

1.Read auth.py to understand the authentication logic and identify potential bugs

2.Read login.py to understand the login flow and identify potential bugs

3.Analyze both files together to identify bugs that span across both files

4.Fix identified bugs in auth.py (syntax, logic, or security)

5.Fix identified bugs in login.py (function calls, syntax)

6.Verify that the fixes are consistent across both files

files: auth.py, login.py

Your codebase has patterns. Your tools should learn them.

Traditional coding assistants start from zero every session. NARE CLI indexes your solutions into persistent semantic memory — turning repeated logic into instant, cached reflexes.

Capabilities

Production-grade reasoning infrastructure. Every feature is designed to minimize token waste and maximize engineering velocity.

O(log N)

lookup time

Semantic Memory

FAISS-powered episodic storage. Every solved task is indexed — sub-100ms similarity search across your entire history.

decision layers

5-Tier Routing

Queries are classified and routed through the optimal path: DIRECT → COMPILED_SKILL → FAST → HYBRID → SLOW.

Auto

skill compilation

Compiled Skills

Recurring patterns crystallize into executable Python modules. The system learns your codebase's idioms.

99.2%

fix accuracy

Verified Synthesis

Generate → test → critique → retry. Solutions are validated before application, with automatic repair on failure.

max iterations

Agent Loop

Autonomous multi-step execution with tool registry, budget control, and extended thinking for complex tasks.

90%

tokens saved

Token Optimization

Prompt caching saves 90%+ on repeated context. File read cache, reasoning cache, and efficient tool workflows.

Installation

Get running in under 60 seconds. No complex setup, no Docker required.

01 — Install

pip install narecli

Requires Python 3.10+

02 — API Key

export ANTHROPIC_API_KEY="sk-..."

Works with Anthropic API or any compatible proxy

03 — Launch

nare

Opens interactive REPL in your project directory

04 — One-Shot

nare "fix the auth bug"

Run a single task without entering the REPL

Alternative — Docker

$ docker-compose up -d

$ docker exec -it nare-cli nare

Five layers of reasoning. One command to rule them all.

From cached skill execution to full verified synthesis with critic feedback — NARE routes every query through the optimal computational path.

Architecture

Five-layer stack from terminal to storage. Built for maximum efficiency, minimum token waste.

CLI Layer

Interactive REPLAgent RendererTheme EngineThinking Display

Agent Layer

Triage AgentCoder AgentPlanning AgentAutonomous Loop

Core Engine

Reasoning RouterVerified SynthesisEvolution EngineCritic / Oracle

Memory & Storage

FAISS IndexEpisodic MemoryCompiled SkillsReasoning Cache

Tools & Execution

File I/OBash / ShellDocker SandboxWeb Search

Routing Decision Table

Route	Tokens	Latency	When
DIRECT	0	<50ms	Greeting or cached conversational response
COMPILED_SKILL	0	<100ms	Execute pre-compiled Python skill module
FAST	0	<100ms	Exact match from episodic memory (FAISS)
HYBRID	~2k	3-8s	Memory context + targeted LLM generation
SLOW	~10k	10-30s	Full verified synthesis with critic + oracle

Reference

Commands, shortcuts, and configuration.

Commands & Shortcuts

/helpShow all available commands

/read <file>Load file into context for the current session

/agentToggle autonomous agent loop on/off

/skillsList compiled skills from memory

/metricsShow performance analytics and route distribution

/historyView conversation history

/clearClear current context and chat history

/modeSwitch between Manual / Autopilot / Agent modes

TabCycle through available modes

Ctrl+LClear terminal screen

Ctrl+DExit NARE CLI

Environment Variables

ANTHROPIC_API_KEY

Your Anthropic API key (required)

Example: sk-ant-...

ANTHROPIC_BASE_URL

Custom proxy URL

Example: https://your-proxy.com

ANTHROPIC_MODEL

Model to use

Example: claude-sonnet-4-20250514

NARE_AGENT_LOOP

Enable agent loop by default

Example: 1

NARE_LOG_LEVEL

Log verbosity

Example: INFO | DEBUG | WARNING

System Requirements

Python 3.10 or higher
Anthropic API key (Claude Sonnet 4 recommended)
512 MB RAM minimum (FAISS index grows with usage)
Windows / macOS / Linux
Optional: Docker for sandboxed code execution

Enterprise Grade

NARE Cloud

The logical evolution of your local CLI. Transition from a workstation script to a distributed, persistent infrastructure for autonomous engineering.

Hosted Agents (24/7 Autonomy)

Global Shared Memory & Skill Synchronization

Compute-as-a-Service (H100/A100 Clusters)

Native CI/CD & Slack/Telegram Integrations

Explore Cloud Infrastructure

Cloud Core Systems Status: Online

The future of AI is not generating more tokens.

Our objective is to shift computational overhead from output tokens to internal hidden states.

Get Started

NARE CLISEMANTICMEMORY