Configuration

Delta works with zero config. Customize when you need more control.

---

Config File

After delta init, configuration is stored at .delta/config.json:

{
  "version": "1.0",
  "budget": {
    "preset": "conservative",
    "maxTokens": 2000,
    "autoEscalate": true
  },
  "graph": {
    "maxDepth": 2,
    "includeTestFiles": true,
    "resolveNodeModules": false
  },
  "relevance": {
    "semanticThreshold": 0.45,
    "embeddingModel": "nomic-embed-text",
    "combineWithGraph": true
  },
  "indexing": {
    "watchMode": false,
    "incrementalDelay": 500
  },
  "embeddings": {
    "provider": "ollama",
    "model": "nomic-embed-text",
    "baseUrl": "http://localhost:11434",
    "dimensions": 768,
    "timeout": 30000
  }
}

---

Token Budget

Presets

Preset	Tokens	Best for
conservative	2,000	Single file changes, quick fixes
balanced	4,000	Feature work, multi-file changes
thorough	8,000	Large refactors, architecture changes

Override per task

delta run "task" --budget 4000
delta run "task" --budget 8000

Auto-Escalation

Delta automatically expands the budget based on the number of changed files:

Changed Files	Budget
< 5 files	Configured budget (no change)
5–9 files	balanced (4,000 tokens)
≥ 10 files	thorough (8,000 tokens)

Disable with:

{
  "budget": {
    "autoEscalate": false
  }
}

---

Embedding Providers

Ollama (Default — Local, Free, Private)

{
  "embeddings": {
    "provider": "ollama",
    "model": "nomic-embed-text",
    "baseUrl": "http://localhost:11434",
    "dimensions": 768,
    "timeout": 30000
  }
}

Setup:

# Install from https://ollama.ai
ollama pull nomic-embed-text
ollama serve

Ollama runs entirely on your machine. No data leaves your computer.

OpenAI

{
  "embeddings": {
    "provider": "openai",
    "model": "text-embedding-3-small",
    "dimensions": 1536
  }
}

Set your API key:

export OPENAI_API_KEY="sk-..."

Azure OpenAI

{
  "embeddings": {
    "provider": "azure",
    "model": "text-embedding-ada-002",
    "dimensions": 1536
  }
}

Set your credentials:

export AZURE_OPENAI_API_KEY="your-key"
export AZURE_OPENAI_ENDPOINT="https://your-resource.openai.azure.com"

LM Studio

{
  "embeddings": {
    "provider": "ollama",
    "model": "nomic-embed-text",
    "baseUrl": "http://localhost:1234"
  }
}

LM Studio uses the same OpenAI-compatible API as Ollama. Just change the baseUrl.

Check status

delta providers

---

Graph Settings

{
  "graph": {
    "maxDepth": 2,
    "includeTestFiles": true,
    "resolveNodeModules": false
  }
}

Setting	Default	Description
maxDepth	2	How deep to trace the dependency graph from changed files
includeTestFiles	true	Whether to include test files in the graph
resolveNodeModules	false	Whether to resolve imports from node_modules

---

Relevance Scoring

{
  "relevance": {
    "semanticThreshold": 0.45,
    "embeddingModel": "nomic-embed-text",
    "combineWithGraph": true
  }
}

Setting	Default	Description
semanticThreshold	0.45	Minimum cosine similarity score to include a file
embeddingModel	nomic-embed-text	Embedding model for semantic scoring
combineWithGraph	true	Combine semantic + graph scores (recommended)

---

.deltaignore

Works like .gitignore. Place at your project root. Delta also inherits your .gitignore automatically.

# .deltaignore
node_modules/**
dist/**
build/**
*.generated.ts
*.min.js
coverage/**
.next/**
__pycache__/**
*.pyc
vendor/**

You don’t need to add node_modules or dist — Delta ignores them by default. Use .deltaignore for project-specific exclusions.

---

Language Support

Tier 1 — Full AST Parsing (tree-sitter)

Language	Extensions	Symbol Extraction
TypeScript	.ts, .tsx	Functions, classes, interfaces, types, imports, exports
JavaScript	.js, .jsx, .mjs	Functions, classes, imports, exports
Python	.py	Functions, classes, imports, decorators
Go	.go	Functions, types, interfaces, imports
Rust	.rs	Functions, structs, traits, impls, mods
Java	.java	Classes, methods, interfaces, imports

Tier 2 — Pattern Extraction (regex)

C, C++, C#, Ruby, PHP, Swift, Kotlin, Scala, Dart, R, Lua, Perl, Haskell, Elixir, Clojure, and more.

Extracts function signatures, class definitions, and imports via regex patterns.

Tier 3 — Notebook Support

Format	Extensions
Jupyter Notebook	.ipynb
Databricks	.dbc

Extracts code cells, markdown cells, and metadata.

Tier 4 — Minimal Indexing

Shell scripts (.sh, .bash, .zsh), config files (.yml, .json, .toml, .xml), markup (.md, .html), styles (.css, .scss, .less), and 20+ additional formats.

Indexed for change detection, dependency tracking, and token counting. No symbol extraction.

56+ file extensions supported across all tiers.

---

Performance Tuning

Setting	Impact
Lower maxDepth	Faster indexing, less context
Higher semanticThreshold	Fewer files included, more precise
resolveNodeModules: false	Much faster indexing (default)
includeTestFiles: false	Reduces index size for large test suites
autoEscalate: false	Consistent budget, predictable costs

---

Monorepo Support

Delta auto-detects monorepo setups:

Tool	Detection
Nx	nx.json
Turborepo	turbo.json
pnpm workspaces	pnpm-workspace.yaml
npm/yarn workspaces	package.json workspaces field

Cross-package imports are resolved automatically:

import { Button } from '@myapp/ui'
// → resolves to packages/ui/src/index.ts