chat-recall

Product Introduction

1. Definition: Chat-recall is a self-hosted, privacy-first AI chat history aggregation and analysis platform. Technically, it is a Node.js-based CLI tool and service that functions as a Model Context Protocol (MCP) server, integrating with AI coding assistants to provide unified memory and security auditing.
2. Core Value Proposition: It exists to solve the problem of fragmented and unsearchable AI conversation history across multiple tools (like Claude Code, Cursor, and others), while proactively identifying and managing security risks like leaked API keys directly within the user's local environment before any sensitive data is transmitted.

Main Features

1. Cross-Platform AI Chat History Aggregation: The tool parses and indexes conversation logs from multiple, incompatible AI coding assistants and chat interfaces. It works by locating local data stores or log files from supported tools, normalizing the data into a unified schema, and making it searchable from a single interface. This process is initiated via the npx chat-recall init command.
2. Local-First Secret Redaction & Security Auditing: A core technical feature is its on-device secret scanning engine. Before any data leaves the user's computer, the tool scans for and redacts known secret patterns (e.g., sk-live-* for OpenAI keys). It then performs a security audit, checking which of the discovered keys are still active and reporting where they were found, providing a critical layer of DevSecOps automation.
3. MCP Server Integration for AI Assistants: Chat-recall exposes the aggregated, redacted history and audit findings to AI assistants via the Model Context Protocol (MCP). This allows assistants like Claude Desktop to directly query the unified history, enabling them to recall past decisions, code snippets, and tasks without requiring the user to manually search and copy-paste context.
4. Centralized Configuration & State Synchronization: The platform provides a dashboard to view and manage the setup across different machines and AI tools. It synchronizes project-specific rules (e.g., prototype vs. production guardrails) and installed add-ons, ensuring a consistent environment and "memory" that follows the user across devices and assistant applications.

Problems Solved

1. Pain Point: Knowledge and context loss due to AI chat history being siloed within individual, proprietary tools that lack cross-compatibility and effective search. This leads to repeated work, forgotten decisions, and difficulty onboarding team members.
2. Pain Point: Unintentional secret leakage in AI conversations, creating a significant security vulnerability that is difficult to track and remediate manually across thousands of messages.
3. Target Audience: Development teams and individual engineers who regularly use multiple AI-powered coding assistants; Engineering and Security leads responsible for code security and developer tooling; Product teams that use AI for planning and documentation.
4. Use Cases: A developer needs to find a specific architecture decision discussed with an AI assistant three months ago. A security audit requires checking if any live API keys have been pasted into chat histories. A team onboarding a new member needs to provide them with the complete history of AI-assisted problem-solving for a project.

Unique Advantages

1. Differentiation: Unlike cloud-based chat history services, chat-recall operates on a strict local-first principle for sensitive data processing. Secrets are redacted before leaving the device, contrasting with services that might ingest full logs. Unlike manual log aggregation, it provides automated, structured analysis and direct integration into the AI workflow via MCP.
2. Key Innovation: Its dual function as both a privacy-preserving search index and a proactive security scanner specifically designed for the new attack surface created by AI assistant chats. The integration via MCP is also innovative, turning the aggregated data into live, queryable context for the very AI systems that generated the data, creating a self-improving loop.

Frequently Asked Questions (FAQ)

1. How does chat-recall handle my sensitive data and API keys? Chat-recall performs all initial processing, including secret pattern detection and redaction, locally on your machine. Only the redacted text (with keys replaced by placeholders like ****) is synchronized to its servers for indexing and search, ensuring raw secrets never leave your control.
2. Which AI coding assistants and tools are compatible with chat-recall? The tool is designed to support multiple AI coding assistants. The documentation specifically mentions compatibility with formats used by tools like Claude Code and Cursor, and it is built to be extensible to other platforms that keep local chat logs.
3. What is the Model Context Protocol (MCP) and how does chat-recall use it? MCP is a protocol developed by Anthropic to allow external tools to provide context to AI assistants safely. Chat-recall acts as an MCP server, giving your AI assistant (like Claude Desktop) the ability to directly search and recall information from your unified chat history, as if it had its own memory.
4. Can chat-recall be used by a team, and how is pricing structured? Yes, chat-recall offers team plans. Pricing includes a free tier for self-hosted individual use, paid monthly/annual subscriptions for solo cloud users, and per-seat pricing for teams, with options for both cloud and self-hosted team deployments.
5. What happens if I switch to a new computer or try a different AI assistant? Once you sign into your chat-recall account on a new device, your entire processed and redacted history, along with your configured project rules and settings, synchronizes automatically. The MCP add-ons ensure any supported AI assistant you use can immediately access the shared history.