Agent♥︎Age
Catalog

Selvedge

Official

by masondelan · Python

Persistent decision memory for AI agents: rationale, rejected approaches, and entity history.

io.github.masondelan/selvedge MCP Server

selvedge (“Decision provenance for AI-coded codebases”) provides the why behind changes in AI-coded projects, including what was already tried and rejected. It focuses on decision provenance, codebase change tracking, and long-term memory concepts, using Model Context Protocol (MCP) tooling.

🛠️ Key Features

  • Decision provenance for AI-coded codebases
  • “Why” for changes, plus what was tried and rejected
  • Codebase change tracking
  • Agent trace and AI agent reasoning
  • MCP, including MCP server usage
  • Git-blame and git-blame-for-ai integration concepts
  • Schema-change-history, stored in sqlite

🚀 Use Cases

  • AI agent–assisted coding with explainable decisions
  • Tracking prior attempts (accepted vs. rejected) in code history
  • Long-term memory for agent context over time
  • Maintaining reasoning context for developer tools workflows

⚡ Developer Benefits

  • Better context for why changes were made
  • Links between AI agents and codebase memory concepts
  • Improved traceability via git-blame-for-ai and change-history

⚠️ Limitations

  • Description provided does not specify supported tools, interfaces, or setup details beyond the decision-provenance and tracking scope.

Topics

agent-memoryagent-traceai-agent-reasoningai-agentsai-code-provenanceai-codebase-memoryai-codingcodebase-change-trackingdecision-memorydeveloper-toolsdevtoolsgit-blamegit-blame-for-ailong-term-memorymcpmcp-servermodel-context-protocolpythonschema-change-historysqlite