Selvedge is a tool that provides "decision memory" for AI coding agents. Its core idea is to retain the reasons for code changes and which options were rejected, so that the "why" persists alongside the code.
Purpose In AI-assisted programming scenarios, agents often see only the final state of the code, without knowing why it was implemented this way originally or which options were ruled out. Selvedge is used to preserve this decision information so that the AI agent in the next session can query and reuse it, thereby avoiding repeated discussions or changes that conflict with past decisions.
Main capabilities
- Provides local decision memory for any compatible AI agent.
- Integrates via either MCP (Model Context Protocol) or CLI.
- Records "what was chosen, what was ruled out, and why."
- Supports looking these records up in subsequent sessions.
How to use
- Use it with a compatible agent; the official prompt lets you paste the specified content into your coding agent to get started.
- Can be invoked via MCP or CLI.
- Specific installation commands and configuration steps are not yet provided.
Applicable scenarios Suitable for developers who use AI coding agents for development and want to preserve architectural and technical decision context across sessions.
Limitations
- Only applies to compatible AI agents; the official guidance suggests checking whether your own agent is compatible.
- Specific support scope, installation details, and compatibility list are not yet provided.
- Free and open source, with data stored locally; beyond this, information on pricing, cost, platforms, or effectiveness is not yet provided.
