What problem does it solve? When you work across multiple AI agents (Claude, Codex, Hermes, OpenClaw, Copilot, Pi), knowledge about how you solved a problem gets trapped in one agent's history. This Skill finds sessions about a specific topic in a chosen agent's raw history, distills them into wiki pages, and returns a synthesized answer you can use immediately in your current session. ## Core Features & Use Cases - Targeted Cross-Agent Search: Score and rank sessions by title match, project directory, and recency decay to find the 3-5 most relevant conversations without reading every file. - Per-Agent Extraction: Uses tailored parsing strategies for each agent's format (Claude JSONL, Codex rollouts, Hermes memories, OpenClaw MEMORY.md, Pi tree-structured sessions) to extract only query-relevant blobs. - Wiki Distillation: Writes or updates categorized wiki pages (skills, concepts, entities, synthesis) with frontmatter, cross-links, and source provenance, then updates the manifest, log, and hot files. - Use Case: You are working in Claude Code and remember solving an auth middleware bug in Codex last month. Run /wiki-codex "auth middleware" to pull that session's knowledge into your wiki and get an immediate synthesized answer. ## Quick Start Ask the agent to run /wiki-codex with the topic "how did I set up auth middleware" to search Codex history and return a synthesized answer.
