Skip to main content

Capabilities

Overview

Lightflare includes several built-in capabilities that work together to run useful agents:

  • Chat provides the interactive conversation experience and streams progress while the agent works.
  • Memory keeps reusable context across chats, users, sessions, and uploaded documents.
  • Workflow runs repeatable multi-step processes from manual starts, trigger records, or chat-created definitions.
  • Agent provides the planning and execution runtime that coordinates memory, skills, tools, checkpoints, replanning, and responses.

Each capability page explains what the feature supports, how it behaves at a high level, and which settings control it.

Properties

Spring Boot supports relaxed binding, so any property can be overridden with an uppercase environment variable where dots and dashes become underscores.

Example:

lightflare.llm.max-output-tokens

becomes:

LIGHTFLARE_LLM_MAX_OUTPUT_TOKENS

Auth

Bootstrap superadmin properties are required only while creating the first superadmin user:

VariableDescription
LIGHTFLARE_BOOTSTRAP_SUPERADMIN_USERNAMEBootstrap superadmin username.
LIGHTFLARE_BOOTSTRAP_SUPERADMIN_EMAILBootstrap superadmin email.
LIGHTFLARE_BOOTSTRAP_SUPERADMIN_PASSWORDBootstrap superadmin password.

Optional auth properties:

VariableDefaultDescription
LIGHTFLARE_AUTH_COOKIE_SECUREfalseSet secure auth cookies.
LIGHTFLARE_AUTH_COOKIE_SAME_SITELaxAuth cookie SameSite mode.

Use the capability pages in this section for chat, memory, workflow, and agent behavior. Use Configurations for runtime properties, and Tools for built-in integrations and MCP.