Claude adds memory import feature to migrate from other AI providers

Claude's memory import feature
Claude has introduced a memory import feature that lets users switch from other AI providers without losing their established context and preferences. The feature is designed to help users who have invested time training other AI systems to bring that knowledge into Claude.
How it works
The import process involves two simple steps:
- Step 1: Copy and paste a provided prompt into a chat with any AI provider. This prompt is specifically written to extract all your context in one chat.
- Step 2: Copy and paste the results into Claude's memory settings. Claude then updates its memory and you can continue conversations where you left off.
The company claims the entire process takes "under a minute" and that your first conversation with Claude will "feel like your hundredth."
Memory features
According to the source material, Claude's memory system:
- Learns your preferences across conversations
- Keeps project context separate to prevent information bleeding between different projects
- Lets you see and edit everything it remembers
Availability and requirements
The memory feature is available on all paid Claude plans. Users need to be on a Pro plan to use the import functionality. The feature is positioned as a way for new Claude users to have their AI "know you from day one."
The source material includes several example prompts that users might give Claude, such as:
- "Help me develop a unique voice for an audience"
- "Improve my writing style"
- "Brainstorm creative ideas"
- "Explain a complex topic simply"
- "Help me make sense of these ideas"
These examples suggest the types of tasks where maintaining context across conversations would be particularly valuable.
📖 Read the full source: HN AI Agents
👀 See Also

Kreuzberg v4.7.0 adds code intelligence for 248 languages and improved markdown extraction
Kreuzberg v4.7.0, a Rust-core document intelligence library, now supports code extraction for 248 formats via tree-sitter and has significantly improved markdown quality with Structural F1 scores over 80% across 23 formats.

SkyClaw Adds Encrypted Chat-Based API Key Setup for AI Agents
SkyClaw implements AES-256-GCM encrypted key ingestion through chat, intercepting key commands at the system layer so the LLM never sees API keys and using one-time key encryption so messaging platforms only see ciphertext.

Exploiting LLM Hidden Agency Signal (Â) for Better Tool Calling
A developer discovered that LLMs have a linearly separable hidden state direction called  that predicts tool calls with AUC > 0.94. Using this signal to force tool calls improved Qwen3-1.7B performance from 26.7% to 85% (+58% gain) and reduced no-tool failures from 43% to 2.6%.

Flue: A TypeScript Framework for Building Autonomous Coding Agents
Flue is a TypeScript framework that provides a programmable harness for building autonomous agents, featuring skills, sessions, sandboxed shell execution, and a built-in virtual sandbox. It can replace tools like Dosu, Greptile, CodeRabbit, Devin, and Claude Code with custom agent logic.