Skillsync enables portable AI chat sessions across coding agents
Open-source tool breaks vendor lock-in by converting agent sessions into interoperable formats
1The vendor lock-in problem
Current AI coding agents store conversations in proprietary formats, trapping users in single platforms. As developers invest time building skills and context within an agent, switching becomes costly. This creates invisible lock-in similar to payment processors or cloud providers. Skillsync addresses this by making sessions portable across different agents like Claude Code, Codex, and Cursor while preserving full conversation history.
2How the conversion works
Skillsync's core is txcript, an open-source Rust engine that translates between agents' storage formats. It maps all session components: conversation history, reasoning steps, and tool calls. The system functions like ffmpeg for video or pandoc for documents, but specialized for AI agent sessions. Everything runs locally except when explicitly sharing to team workspaces, maintaining privacy and control over sensitive coding conversations.
3Session management features
The desktop app aggregates scattered agent sessions into a unified interface. It makes normally unreadable binary formats human-readable and searchable. Users can track which skills each session utilizes and identify stale context. Shared workspaces enable teams to sync sessions, allowing collaborators to build on each other's work rather than duplicate efforts. Skills and memories are stored as portable markdown files.
4Real-world use cases
Developers use Skillsync to continue sessions when hitting usage limits on one agent, switch between specialized agents mid-task, or hand off browser-based research to code-generation tools. Teams aggregate research sessions so members can leverage collective work. The system enables non-technical collaborators to pick up where engineers left off, making AI sessions persistent collaboration units rather than disposable logs.
5Technical implementation
Skillsync exposes portable skills and memories to any agent via MCP (Memory Context Protocol) for search and selective retrieval. The local-first architecture ensures sensitive coding conversations remain on-device unless shared. The team's prior experience building an open-source payment orchestrator with 30k+ GitHub stars informed their approach to solving interoperability challenges in the AI agent space.
6Current adoption
Available as a Mac app, CLI tool, and via MCP integration, Skillsync has demonstrated value in Y Combinator's W26 batch. Early adopters report successfully moving sessions between Claude Code, Codex, and Cursor while preserving context. The solution particularly benefits engineering teams collaborating on complex problems where different agents excel at distinct phases of the workflow.
7Industry implications
Skillsync challenges the emerging paradigm where AI tool vendors benefit from proprietary session formats. By making context portable, it shifts power dynamics in the agent ecosystem. This approach could influence how future AI tools handle interoperability, similar to how open standards transformed other developer tools. The solution reveals how much valuable work currently gets discarded with each new session.
8Future potential
As AI agents become more specialized, portable sessions will enable composite workflows leveraging multiple tools. Skillsync's architecture suggests possibilities for standardized agent communication protocols. The ability to analyze patterns across sessions could lead to better workflow insights. By treating sessions as persistent artifacts, developers may begin versioning and refining them like code.