haft
View on GitHubEngineering decisions engine that know when they're stale. Frame, compare, decide — with evidence decay and parity enforcement. For Claude Code, Cursor, Gemini CLI, Codex and more.
Haft adds local project memory and engineering-decision governance to coding agents. It provides MCP integrations and skills for framing problems, comparing options, recording decisions, and checking evidence or specifications for staleness.
Use Cases
Keep durable, structured project memory for coding agentsFrame engineering problems and acceptance criteriaCompare solution options with evidence and parity rulesRecord and review binding engineering decisionsDetect stale evidence and specification driftIntegrate governance skills and MCP tools with coding-agent hosts
Built With
- Language
- Go
- Frameworks
- First Principles Framework (FPF) · Model Context Protocol (MCP)
Tags
AI coding agents · project memory · engineering decisions · MCP · agent skills · evidence tracking · decision staleness · spec-driven development · local-first