AgentRecall: local-first memory for AI coding agents in .NET

[AkbarDizaji]

Pre-submission Checklist

• This post follows the MCP community guidelines

What would you like to share?

Hi everyone,

I built AgentRecall, a local-first memory and learning tool for AI coding agents, built as a .NET global tool and NuGet package.

It turns feedback, failures, PR review comments, build/test/lint logs, and repeated mistakes into structured reusable rules. Then it ranks the relevant rules and injects them back into the agent context automatically through a Claude Code/MCP workflow.

How it is different from built-in memory in tools like Claude:

• It is developer-controlled and project/tool-level, not tied to one chat product
• Memory stays local: no cloud sync, no web UI, no API keys by default
• Rules are scored, ranked, promoted, archived, or superseded instead of being stored as plain notes
• Relevant rules are injected deterministically into context, not only when the model “remembers” to ask
• Repeated failures can increase rule confidence and help the agent avoid the same mistake next time

A few things it supports:

• Must-follow, suggested, and warning rules within a token budget
• Standing rules that apply every turn
• Automatic conflict resolution between rules
• Learning from build/test/lint failures and PR comments
• Seed packs like tidy-first and career-impact to give a new project useful starting guidance
• Project DNA summaries for onboarding agents or developers quickly

I would love feedback from the MCP community on whether this kind of local rule-memory layer fits MCP workflows, and what examples would make it easier to try.

Relevant Links

• GitHub repository: https://github.com/AkbarDizaji/AgentRecall
• NuGet package: https://www.nuget.org/packages/AgentRecall
• Author profile: https://github.com/AkbarDizaji