Introducing the Agent Development Environment Aging

Published: January 15, 2025 | Original Post

---

A visual development platform making AI agent design and debugging transparent.

Key Features

State and Context Management

The ADE provides developer control over:

• System prompts (unchanging agent instructions)
• Tool definitions and their inputs/outputs
• Core memory (editable in-context storage)
• Character limits for memory and tool responses

Memory Architecture

In-Context Memory (Core Memory): Persistent, editable blocks that agents can update via core_memory_append and core_memory_replace tools. Most implementations use two blocks: one for human-related memories and another for agent persona.

External Memory: Includes recall memory (searchable conversation history) and archival memory (vector database storage accessible through archival_memory_search).

Transparent Reasoning

Letta requires all agents to show their work, regardless of whether they are powered by Gemini, OpenAI, or Anthropic. Agents must explicitly use the send_message() tool to communicate with users.

Tool Integration

Built-in tools plus 7,000+ pre-built tools via Composio. Custom tools can be created using an embedded Python editor with real-time testing.

Deployment Modes

Mode	Description
Debug	Full visibility
Interactive	Balanced visibility and usability
Simple	End-user perspective