letta-ai / claude-subconscious

2026-01-14 · Source: Github Trending: All languages · Field: Technology & Digital — Software Development & Engineering, Artificial Intelligence & Machine Learning · Depth: Intermediate, long

Summary

Claude Subconscious is an experimental background agent designed to extend Claude Code's capabilities by providing persistent memory, tool access, and context engineering. It operates by watching Claude Code session transcripts, reading the user's codebase using tools like Read, Grep, and Glob, and remembering information across sessions and projects. This agent then "whispers" guidance, context, patterns, and reminders before each prompt, without blocking the user's workflow. Unlike Claude Code, which forgets information between sessions, Claude Subconscious leverages Letta's memory system and SDK to build a cumulative understanding of the user's work, offering features like persistent project context, learned preferences, and background research. Installation involves adding it as a plugin to Claude Code, with various configuration options for API keys, agent modes (whisper, full, off), and client-side tool access (read-only, full, off).

Key takeaway

For AI Engineers and Machine Learning Engineers using Claude Code, integrating Claude Subconscious can significantly enhance your development workflow by providing persistent project context and proactive guidance. This allows your coding agent to learn from past interactions and codebase specifics, reducing repetitive tasks and improving efficiency. Consider configuring `LETTA_SDK_TOOLS` to `read-only` initially for safe background research, then explore `full` mode for advanced automation as you gain confidence in the agent's capabilities.

Key insights

Claude Subconscious enhances Claude Code with persistent memory and tool-enabled contextual guidance.

Principles

• Persistent memory improves agent utility.
• Background agents can provide non-intrusive guidance.
• Tool access expands an agent's operational scope.

Method

A background Letta agent watches Claude Code sessions, reads files, updates memory, and whispers guidance via stdout before prompts, using an async hook pattern to avoid blocking.

In practice

• Use `LETTA_MODE="full"` for comprehensive context injection.
• Configure `LETTA_SDK_TOOLS="full"` for agent autonomy.
• Set `LETTA_MODEL` to `anthropic/claude-sonnet-4-5` for better reasoning.

Topics

• Claude Subconscious
• Letta Agents
• Persistent Memory
• AI Code Assistants
• Tool-Augmented AI

Code references

• letta-ai/letta-code
• letta-ai/claude-subconscious
• letta-ai/claude-subconscious
• anthropics/claude-code

Best for: AI Architect, NLP Engineer, AI Engineer, Machine Learning Engineer, Software Engineer

Related on AIssential

• Give Claude Permanent Memory — Claude's memory architecture offers layered strategies for persistent context, moving beyond stateless interactions.
• OpenAI Has Some Catching Up to Do — Claude Code's "vibe coding" approach is gaining traction over OpenAI's more structured, senior-engineer-focused Codex.
• I Let Claude Code for Me for 30 Days. Here’s What Nobody’s Telling You. — Extensive AI coding assistance can degrade developer instincts and deep code comprehension through cognitive offloading.
• I Use Claude Code Every Day. Here's the Setup That Actually Matters — Claude Code is an agentic CLI for AI-assisted development, not just an autocomplete tool.
• Claude Code Just Got A MASSIVE Upgrade! New Agent Skills, Ultra Think, Voice Mode, & More! — Claude Code updates enhance AI agent capabilities through improved skill creation, extended reasoning, and remote access.
• Comment remplacer ChatGPT par Claude (et pourquoi) — Claude's agentic AI model, with persistent memory and file interaction, offers a strategic advantage over traditional chatbots.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Github Trending: All languages.