Does it store simple things like basic syntax?

No, the skill is configured to skip obvious, Google-able facts and focus exclusively on complex, non-obvious lessons and unique architectural patterns.

How does the retrieval process work?

It uses a two-stage process: first searching previews for semantic matches via an MCP server, then analyzing similarities to decide whether to merge, update, or create new entries.

What triggers a memory to be stored?

Memories are stored manually using the --store flag, or automatically when Claude detects a difficult task completion or strong user frustration signals.

Why does it use a sub-agent for storage?

It uses a 'general-purpose' sub-agent via the Task tool to keep the main conversation context clean and focused on the primary development task.

Can I categorize my stored coding patterns?

Yes, the skill automatically detects roles such as backend, frontend, devops, and quant based on the task context to organize memories into specific collections.

Coder Memory Store

Name: Coder Memory Store
Author: hongbietcode

byhongbietcode

•

开发者工具

Captures and retrieves hard-won coding patterns and architectural lessons in a persistent vector database.

关于

Coder Memory Store transforms Claude Code into a self-improving development partner by building a long-term repository of complex problem-solving patterns. Unlike standard documentation, this skill specifically targets non-obvious bugs, architectural insights, and 'hard lessons' that are often lost between sessions. It uses a sophisticated two-stage retrieval process via MCP server v2 to intelligently merge new insights with existing knowledge, ensuring your AI assistant learns your specific preferences and avoids repeating past mistakes across different projects.

主要功能

Intelligent role detection for organized knowledge across domains like Backend, AI, and DevOps
Automated pattern extraction from complex task successes or failures
Vector database storage for semantic search and retrieval of coding insights
Multi-stage consolidation logic to merge, update, or generalize learned patterns
2 GitHub stars
Frustration-triggered learning to capture critical context during difficult debugging sessions

使用场景

Preserving 'gotchas' and non-obvious bug fixes for reuse in future development cycles
Building a personalized architectural pattern library that evolves with your coding style
Maintaining continuity on complex projects by recalling past implementation logic and trade-offs

关于

主要功能

Intelligent role detection for organized knowledge across domains like Backend, AI, and DevOps
Automated pattern extraction from complex task successes or failures
Vector database storage for semantic search and retrieval of coding insights
Multi-stage consolidation logic to merge, update, or generalize learned patterns
2 GitHub stars
Frustration-triggered learning to capture critical context during difficult debugging sessions

使用场景

Preserving 'gotchas' and non-obvious bug fixes for reuse in future development cycles
Building a personalized architectural pattern library that evolves with your coding style
Maintaining continuity on complex projects by recalling past implementation logic and trade-offs