Pulls specific knowledge from one AI agent's conversation history into your current session with another agent. You're working in Claude and need to remember how you solved auth middleware in Codex last month? Run `/wiki-codex auth middleware` and it scores sessions by relevance, extracts just the pertinent conversation windows (not everything), and drops a synthesized answer you can use immediately. Works across Claude, Codex, Hermes, OpenClaw, and Copilot histories. The scoring is simple but smart: weighs title matches, project context, and recency without opening full session files until it knows what matters. Different from bulk archival tools because this is cross-referencing on demand, not building an index. Saves you from manually digging through JSONL files or trying to remember which agent you were using when you figured something out.
npx skills add https://github.com/ar9av/obsidian-wiki --skill wiki-agent