Skip to content

ArqonShip Configuration Reference

Configuration File

ArqonShip uses .arqon/config.toml for project-level settings.

Generating Default Config

arqon init

Creates .arqon/config.toml with defaults.

Full Configuration Schema

[meta]
# Configuration version (for migrations)
config_version = 1

[oracle]
# Glob patterns for files to include in scanning
include_globs = [
    "src/**/*.rs",
    "src/**/*.py",
    "lib/**/*.rs",
]

# Glob patterns for files/directories to exclude
exclude_globs = [
    "target/",
    "node_modules/",
    "venv/",
    ".git/",
    "*.min.js",
]

# Path to store/cache model files
model_path = "~/.arqon/models/"

[heal]
# Maximum repair attempts before giving up (Constitution XVII.1)
max_attempts = 2

# LLM model identifier for local inference
model_id = "deepseek-coder-1.3b-instruct"

# Enable/disable self-healing feature
enabled = true

[ship]
# Branches allowed for release (Constitution XVIII)
require_branches = ["main", "release/*"]

# Version scheme: "semver" or "calver"
version_scheme = "semver"

Configuration Options

[meta] Section

Option Type Default Description
config_version integer 1 Schema version for future migrations

[oracle] Section

Option Type Default Description
include_globs string[] ["src/**/*.rs", "src/**/*.py"] Files to parse and index
exclude_globs string[] ["target/", "venv/", ".git/"] Paths to skip
model_path string "~/.arqon/models/" Model cache directory

[heal] Section

Option Type Default Description
max_attempts integer 2 Max healing iterations per failure
model_id string "deepseek-coder-1.3b-instruct" Local LLM model
enabled boolean true Feature toggle

[ship] Section

Option Type Default Description
require_branches string[] ["main"] Branches allowed for shipping
version_scheme string "semver" Versioning strategy

Environment Variables

Variable Description Required
GITHUB_TOKEN GitHub API token for PR creation For arqon ship
ARQON_LLM_URL LLM API endpoint (OpenAI/Ollama compatible) For arqon heal
ARQON_LLM_MODEL Model name (default: deepseek-coder:1.3b) No
ARQON_LLM_KEY API key for LLM endpoint If required by API
ARQON_CONFIG Override config file path No
ARQON_MODEL_CACHE Override model cache directory No

Example LLM Configuration

# For local Ollama
export ARQON_LLM_URL="http://localhost:11434/v1"
export ARQON_LLM_MODEL="deepseek-coder:1.3b"

# For OpenAI
export ARQON_LLM_URL="https://api.openai.com/v1"
export ARQON_LLM_MODEL="gpt-4o-mini"
export ARQON_LLM_KEY="sk-..."

Data Directories

Path Purpose
.arqon/config.toml Project configuration
.arqon/graph.db SQLite graph database
.arqon/vectors.lance/ LanceDB vector storage
~/.arqon/models/ Cached AI models
~/.arqon/audit.db Global audit log

Example Configurations

Minimal (Rust-only project)

[meta]
config_version = 1

[oracle]
include_globs = ["src/**/*.rs"]

[heal]
enabled = false  # Manual review only

Full-stack project

[meta]
config_version = 1

[oracle]
include_globs = [
    "backend/**/*.rs",
    "frontend/**/*.ts",
    "scripts/**/*.py",
]
exclude_globs = [
    "target/",
    "node_modules/",
    "dist/",
    "*.test.ts",
]

[heal]
max_attempts = 3
enabled = true

[ship]
require_branches = ["main", "release/*"]