Dev-Mem AI Programming Experience Knowledge Base

Problem

In AI-assisted coding, debugging conclusions, version constraints, API limits, and prompt patterns often remain scattered across chat windows. When you open a new session, switch tools, or return later, you may re-explain context or repeat the same mistakes. This skill turns those conversational learnings into a searchable local knowledge base instead of relying on memory.

How It Works

• Triggering and routing: It supports @emem as an explicit trigger, as well as natural language such as “save this”, “look it up”, and “review this session”, then routes to workflows for capture, search, preflight checks, daily consolidation, stats, or backup.
• Structured capture: It extracts symptoms, constraints, debugging trail, root cause, fix, conclusion, and caveats from context, then writes them to dev-mem.md in the project root or an environment-specific path without requiring the user to restate the full issue.
• Main workflows:
• Instant capture: Turn a single pitfall or constraint into a structured entry.
• Preflight checks: Before starting a feature, surface relevant historical learnings.
• Search: Retrieve similar past issues by precise query or loose probing.
• Review and anchors: Use mark to flag key nodes during long sessions, then prioritize them during review.
• Stats and repair: Inspect entry counts and fix format, category, or stats inconsistencies.

Boundaries

It operates mainly on the current session context and cannot read earlier chat windows. The storage path changes across IDEs, Knot, Claude Code, and other environments, and preflight relevance depends on how many entries already exist. It fits lightweight personal development knowledge capture rather than replacing team documentation systems or CI/CD processes.