chat-recall logo

chat-recall

Freemium

Собирает транскрипты из ИИ-инструментов для программирования в единую историю с поиском и удаляет учетные данные перед синхронизацией. Помогает разработчикам и командам продолжать работу, изучать прошлые решения, находить проблемы безопасности и делиться контекстом проекта между ассистентами.

Что такое chat-recall?

chat-recall is a memory and search layer for AI coding tools. It reads the conversations, plans, task lists, notes, and shell history that those tools already store, indexes them into one searchable history, and lets assistants retrieve relevant context through an MCP server. It is intended for developers and teams that work across multiple AI coding assistants and need to resume work without manually rebuilding context.

Setup starts from the terminal with npx chat-recall init. The CLI finds supported AI coding tools on the machine, configures their MCP connections, and reads existing transcripts. Credential redaction happens on the user's hardware before data is sent for indexing; the source describes support for standard cloud keys, authentication tokens, private keys, connection strings, and custom patterns.

Что умеет chat-recall?

Cross-tool transcript search

Search sessions from multiple AI coding tools in one ranked result list instead of checking each assistant separately.

Assistant-driven recall through MCP

The MCP server exposes tools for searching sessions, resuming work, inspecting context, querying the knowledge graph, and reading security findings inside the assistant conversation. A fresh installation registers 28 tools by default, with 63 available through the full profile.

Project knowledge and decision history

The index can return project findings, dependencies, risks, decisions, and changes over time, helping users trace why a technical choice was made.

Security finding and secret monitoring

chat-recall detects supported credentials in historical sessions, reports whether a key was found and whether it still works, and groups occurrences across conversations. Users can add custom detection patterns.

Task and work continuity

Findings can become tasks with a proposed fix, while task cards retain the originating session, changed files, commits, outcome, and any point where the assistant ran out of context.

Local synchronization controls

CLI rules can exclude projects, tools, or source folders; create allowlists; delete remote sessions; and inspect active rules. Project settings, skills, and MCP configuration can also be distributed across team machines on supported plans.

Варианты использования

“Resume work on another assistant”

A developer can search prior sessions or use assistant recall to continue a partially completed task in a different AI coding tool without restating the earlier reasoning.

“Onboard a new machine”

After signing in, a developer can access the indexed history and configured add-ons without manually copying transcripts between computers.

“Investigate leaked credentials”

A team can review detected keys in its AI history, see occurrences and current validity, and apply exclusions or deletion rules when particular sources should not sync.

“Turn findings into an action plan”

Developers can review ranked code findings and task cards that link back to the conversation, files, and commits responsible for the work.

“Coordinate shared project work”

Teams can keep shared project history, assign work, use a task board, and control which projects remain private until explicitly shared.

Часто задаваемые вопросы

How do I set up chat-recall?

Run `npx chat-recall init` from the terminal. The CLI indexes existing transcripts and configures supported AI coding tools for MCP access.

What data is sent away from my machine?

The CLI redacts credentials before synchronization, and the raw transcript is processed locally before the indexing stage. Users can exclude projects, tools, or folders, create allowlists, decline sources, or delete synced sessions through the CLI.

Can an assistant search chat-recall without leaving the conversation?

Yes. Its MCP server lets an assistant call recall, search, knowledge-graph, and security tools from within the conversation. The assistant can retrieve relevant past sessions and project context when needed.

Is there a self-hosted option?

Yes. The pricing page lists a free self-hosted tier for one person, including the full feature set, and a self-hosted team option. Hosted Solo and Team plans are also available, with a seven-day trial that does not require a card.

Does chat-recall train on user sessions?

The pricing page states that chat-recall does not train on user sessions.

Краткие сведения

Category
Developer Tool
Primary users
Developers and software teams using AI coding assistants
Setup
Terminal CLI with `npx chat-recall init`
Integration model
MCP server for AI assistant access
Deployment
Self-hosted or hosted options
Source domain
chatrecall.dev

Анализ трафика chat-recall

Данные о трафике приведены только для справки.

Рейтинг домена
16