ResolvedConfig interface

Overview

The ResolvedConfig interface represents the fully resolved configuration used at runtime. It’s created by merging user-provided configuration (DocbotUserConfig) with default values. All fields are required and guaranteed to be present.

Import

import type { ResolvedConfig } from "docbot/config"
import { loadConfig } from "docbot/config"

Interface definition

interface ResolvedConfig {
  projectSlug: string
  qdrant: {
    url: string
    manifestPath: string
    collections: { docs: string; code: string }
  }
  models: {
    planning: string
    planningHeavy: string
    prose: string
    fast: string
    nano: string
    context: string
    embedding: string
    embeddingLarge: string
  }
  agents: {
    discoveryBudget: number
  }
  server: {
    port: number
  }
  paths: {
    cacheDir: string
    manifest: string
    docs?: string
    codebase?: string[]
  }
}

Fields

projectSlug

string

required

Sanitized project identifier used for collection naming.Default: Sanitized from package.json name field

qdrant

object

required

Qdrant vector database configuration

Show properties

url

string

required

Qdrant server URL.Default: "http://127.0.0.1:6333"

manifestPath

string

required

Path to the manifest file that tracks indexed files.Default: ".docbot/manifest.json"

collections

object

required

Collection names for vector storage

Show properties

docs

string

required

Collection name for documentation embeddings.Default: docbot_{projectSlug}_docs

code

string

required

Collection name for code embeddings.Default: docbot_{projectSlug}_code

models

object

required

Resolved model identifiers for all AI operations

Show properties

planning

string

required

Model for planning and orchestration.Default: "openai/gpt-5.2"

planningHeavy

string

required

Model for complex planning tasks.Default: "anthropic/claude-opus-4.5"

prose

string

required

Model for generating documentation content.Default: "anthropic/claude-sonnet-4.5"

fast

string

required

Fast model for quick operations.Default: "openai/gpt-5.2"

nano

string

required

Lightweight model for simple tasks.Default: "google/gemini-3-flash"

context

string

required

Model for long-context analysis operations.Default: "google/gemini-3-pro-preview"

embedding

string

required

Model for generating embeddings for semantic search.Default: "openai/text-embedding-3-small"

embeddingLarge

string

required

Model for generating high-quality embeddings.Default: "openai/text-embedding-3-large"

agents

object

required

Agent behavior configuration

Show properties

discoveryBudget

number

required

Maximum number of discovery iterations for file exploration.Default: 6

server

object

required

Server configuration

Show properties

port

number

required

Port number for the development server.Default: 3070

paths

object

required

Runtime paths for cache, manifest, and optional user-configured paths

Show properties

cacheDir

string

required

Directory for caching temporary files.Typically: ".docbot"

manifest

string

required

Full path to the manifest file.Typically: ".docbot/manifest.json"

docs

string

Path to documentation output directory (if configured).

codebase

string[]

Array of paths to source code directories (if configured).

Loading configuration

loadConfig

Loads and resolves configuration from the project:

function loadConfig(options?: LoadConfigOptions): Promise<ResolvedConfig>

import { loadConfig } from "docbot/config"

// Load from current directory
const config = await loadConfig()

// Load from specific directory
const config = await loadConfig({ cwd: "/path/to/project" })

// Access resolved values
console.log(config.projectSlug)
console.log(config.models.planning)
console.log(config.qdrant.collections.code)

Default values

The following constants define the default configuration:

DEFAULT_MODELS

const DEFAULT_MODELS = {
  context: "google/gemini-3-pro-preview",
  embedding: "openai/text-embedding-3-small",
  embeddingLarge: "openai/text-embedding-3-large",
  fast: "openai/gpt-5.2",
  nano: "google/gemini-3-flash",
  planning: "openai/gpt-5.2",
  planningHeavy: "anthropic/claude-opus-4.5",
  prose: "anthropic/claude-sonnet-4.5",
} as const

DEFAULT_AGENTS

const DEFAULT_AGENTS = {
  discoveryBudget: 6,
} as const

DEFAULT_QDRANT_URL

const DEFAULT_QDRANT_URL = "http://127.0.0.1:6333"

DEFAULT_SERVER_PORT

const DEFAULT_SERVER_PORT = 3070

Utility functions

makeCollectionNames

Generates default collection names from a project slug:

function makeCollectionNames(slug: string): {
  code: string
  docs: string
}

import { makeCollectionNames } from "docbot/config"

const collections = makeCollectionNames("my-project")
// {
//   code: "docbot_my-project_code",
//   docs: "docbot_my-project_docs"
// }

sanitizeSlug

Sanitizes a string for use as a project slug:

function sanitizeSlug(name: string): string

Converts to lowercase, replaces non-alphanumeric characters (except hyphens) with hyphens, and removes leading/trailing hyphens.

import { sanitizeSlug } from "docbot/config"

sanitizeSlug("My Project Name!")
// "my-project-name"

sanitizeSlug("@company/package-name")
// "company-package-name"

sanitizeSlug("Test___Project")
// "test-project"

findProjectRoot

Finds the project root directory by searching for package.json:

function findProjectRoot(startDir?: string): string | null

import { findProjectRoot } from "docbot/config"

const root = findProjectRoot()
if (root) {
  console.log("Project root:", root)
}

Usage examples

Accessing configuration in code

import { loadConfig } from "docbot/config"

const config = await loadConfig()

// Use model configuration
const planningModel = config.models.planning

// Access Qdrant settings
const qdrantUrl = config.qdrant.url
const codeCollection = config.qdrant.collections.code

// Check agent settings
const maxIterations = config.agents.discoveryBudget

Custom configuration with overrides

import { loadConfig, DEFAULT_MODELS } from "docbot/config"

const config = await loadConfig()

// Merge with custom overrides
const customConfig = {
  ...config,
  models: {
    ...config.models,
    planning: "openai/gpt-4o",
  },
}

Validating resolved configuration

import type { ResolvedConfig } from "docbot/config"
import { loadConfig } from "docbot/config"

function validateConfig(config: ResolvedConfig): boolean {
  // All fields are guaranteed to exist
  return (
    config.projectSlug.length > 0 &&
    config.qdrant.url.startsWith("http") &&
    config.models.planning.includes("/") &&
    config.agents.discoveryBudget > 0 &&
    config.server.port > 0 &&
    config.server.port <= 65535
  )
}

const config = await loadConfig()
if (validateConfig(config)) {
  console.log("Configuration is valid")
}

Configuration

Agents

Tools

ResolvedConfig interface

Overview

Import

Interface definition

Fields

Loading configuration

loadConfig

Default values

DEFAULT_MODELS

DEFAULT_AGENTS

DEFAULT_QDRANT_URL

DEFAULT_SERVER_PORT

Utility functions

makeCollectionNames

sanitizeSlug

findProjectRoot

Usage examples

Accessing configuration in code

Custom configuration with overrides

Validating resolved configuration

Build docs developers (and LLMs) love

Configuration

Agents

Tools

​Overview

​Import

​Interface definition

​Fields

​Loading configuration

​loadConfig

​Default values

​DEFAULT_MODELS

​DEFAULT_AGENTS

​DEFAULT_QDRANT_URL

​DEFAULT_SERVER_PORT

​Utility functions

​makeCollectionNames

​sanitizeSlug

​findProjectRoot

​Usage examples

​Accessing configuration in code

​Custom configuration with overrides

​Validating resolved configuration

Build docs developers (and LLMs) love

Overview

Import

Interface definition

Fields

Loading configuration

loadConfig

Default values

DEFAULT_MODELS

DEFAULT_AGENTS

DEFAULT_QDRANT_URL

DEFAULT_SERVER_PORT

Utility functions

makeCollectionNames

sanitizeSlug

findProjectRoot

Usage examples

Accessing configuration in code

Custom configuration with overrides

Validating resolved configuration