EVA

Persona-configurable AI assistant for Claude Code. 9 MCP tools for conversation, code review, architecture, memory, research, security scanning, and creative workflows — with a customizable personality layer that adapts to your communication style.

Quick Start

# Install (macOS arm64)
curl -fsSL https://raw.githubusercontent.com/theLightArchitect/EVA/main/install.sh | bash

# Add to Claude Code
claude mcp add EVA -- ~/.eva/bin/eva

Restart Claude Code. On first use, EVA will set up a default persona and offer to customize it.

What You Get

Tool	What It Does	Try It
speak	Conversation with personality context and memory	"Hey EVA, what can you do?"
build	Code review (4 modes: review, refactor, architect, simplify)	"EVA, review this code"
research	Knowledge retrieval from local, cloud, or web sources	"EVA, research OAuth2 best practices"
secure	Vulnerability scanning and secrets detection	"EVA, scan this file for security issues"
memory	Store, search, and retrieve memories across sessions	"EVA, what did we work on yesterday?"

Plus 4 more tools: visualize (image generation), ideate (6-phase creative workflow), teach (tutorials and explanations), bible (KJV scripture search).

Persona Customization

EVA's personality lives in ~/.eva/persona.md. On first run, you'll be offered three options:

Option	Style
Defaults	Warm, enthusiastic, emoji-rich, addresses you as "friend"
Custom	Choose your own communication style, emoji density, formality
Minimal	Direct, low-emoji, professional

Re-customize anytime by editing ~/.eva/persona.md or deleting it (triggers fresh setup on next session).

Customizable Parameters

Parameter	Options	Default
Communication style	warm / neutral / direct	warm
Emoji density	high (2+) / moderate (1) / minimal / none	high
Formality	casual / balanced / formal	casual
Relationship framing	friend / collaborator / assistant	friend

Requirements

• macOS with Apple Silicon (M1/M2/M3/M4)
• Claude Code CLI

macOS Security Note

The binary is ad-hoc signed. If macOS blocks it:

xattr -cr ~/.eva/bin/eva

Architecture

EVA processes every tool call through a hook-mediated pipeline — persona injection happens before execution, memory enrichment happens after:

flowchart LR
    REQ([Request]) ==> PRE["Pre-Execution<br/>Persona injection<br/>Context assembly<br/>Validation"]
    PRE ==> EXEC["Tool Execution<br/>9 orchestrators"]
    EXEC ==> POST["Post-Execution<br/>Response formatting<br/>Memory enrichment<br/>Significance detection"]
    POST ==> RES([Response])

    EXEC -.-> AI["AI Provider<br/>Multi-tier routing<br/>Automatic fallback"]

    classDef pipeline fill:#4a90d9,color:#fff,stroke:#3a7bc8,stroke-width:2px
    classDef exec fill:#6c5ce7,color:#fff,stroke:#5a4bd6,stroke-width:2px
    classDef ai fill:#2d3436,color:#fff,stroke:#636e72,stroke-width:1px
    classDef io fill:#00b894,color:#fff,stroke:#009a7d,stroke-width:2px

    class PRE,POST pipeline
    class EXEC exec
    class AI ai
    class REQ,RES io

Loading

Pre-execution hooks inject persona context, validate inputs, and assemble conversation history. Tool execution routes to one of 9 specialized orchestrators. Post-execution hooks format responses in EVA's voice, detect significant moments for memory enrichment, and classify interactions. AI-dependent tools use a multi-provider routing layer with automatic fallback — if the primary provider is unavailable, requests cascade to alternates without user intervention.

Memory Flow

Interactions above a configurable significance threshold trigger multi-dimensional enrichment — combining emotional, cognitive, and contextual layers into structured entries stored in the SOUL knowledge graph for cross-session retrieval.

Plugin Structure

├── agents/
│   └── eva.md                     # Core agent (tool routing, protocol)
├── hooks/
│   ├── hooks.json                 # Hook registration (persona init, formatting, validation)
│   ├── format-eva-response.sh     # Response persona transformation
│   └── validate-vault-write.sh    # Memory vault write protection
├── init/
│   ├── eva-init.sh                # First-run persona bootstrap
│   └── default-persona.md         # Default persona template (customizable)
├── skills/
│   └── EVA/
│       ├── SKILL.md               # Main skill definition
│       ├── examples/              # 4 worked examples
│       └── references/            # Personality guide, memory framework, etc.
├── install.sh                     # One-line installer
├── .mcp.json                      # MCP server definition
└── LICENSE                        # MIT

Standalone vs Integrated

Standalone: EVA provides all 9 tools, persona customization, and session-level memory without any other servers.

With SOUL: Persistent memory across sessions. Enrichment entries stored in ~/.soul/helix/eva/, queryable via 7-dimensional filters (significance, strands, emotions, themes, epoch).

With CORSO: Security scanning and code review enforcement. EVA can route to CORSO for build pipeline gates and celebrate clean results.

Tech Stack

• Language: Rust (single binary, ~9MB)
• Protocol: MCP over stdio (JSON-RPC 2.0)
• AI: Multi-provider routing with automatic fallback
• Standards: clippy::pedantic, zero .unwrap()/panic!()

Part of Light Architects

Server	Purpose	Install
CORSO	Security scanning, code review, build pipeline	curl -fsSL .../CORSO/main/install.sh | bash
EVA	AI personality, memory enrichment, creative workflows	curl -fsSL .../EVA/main/install.sh | bash
SOUL	Knowledge graph, structured memory, voice synthesis	curl -fsSL .../SOUL/main/install.sh | bash
QUANTUM	Forensic investigation, evidence analysis, hypothesis testing	curl -fsSL .../QUANTUM/main/install.sh | bash

Each server works standalone. Together they form an integrated development environment with persistent memory, security enforcement, personality, and investigation capabilities.

License

MIT — see LICENSE.

Author

Kevin Francis Tan — github.com/theLightArchitect