菜单
Read-only Meta Knowledge Graph memory lookup for Codex. Use to fetch and summarize relevant project-scoped and user-scoped Learning or Decision context from Neo4j before planning, reviewing, or orchestrating work.
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
基于 SOC 职业分类
Run memory-grounded multiagent orchestration with Meta Knowledge Graph in Codex. Use when the user asks to execute a task through MKG, recall prior learnings first, route memory into subagents, verify phases, and capture durable learnings or decisions.
Onboard or check a Meta Knowledge Graph session in Codex. Use when the user asks to start MKG, verify mounted MKG MCP tools, inspect graph state, choose between the RoadFlex sales demo and a custom persona, or run the Codex equivalent of the Claude Code /mkg-start command.
Set up or verify the MKG RoadFlex sales-agent demo in Codex. Use when the user asks for the sales demo, RoadFlex, demo seeding, Neo4j/BigQuery/Neocarta/Diffbot setup, or the Codex equivalent of the Claude Code sales_agent_demo command.
| name | mkg-recall |
| description | Read-only Meta Knowledge Graph memory lookup for Codex. Use to fetch and summarize relevant project-scoped and user-scoped Learning or Decision context from Neo4j before planning, reviewing, or orchestrating work. |
Retrieve relevant memory from the Meta Knowledge Graph and return only the
useful facts. This is the Codex skill equivalent of the Claude Code
mkg-recall subagent.
project_get_context for ranked recall;
use neo4j_read_cypher for custom read-only queries.approved learnings as policy. Treat candidate learnings as hints
and Decision nodes as context, not policy.toString(...) so
date/datetime values do not serialize as {}.Return a short ranked list of relevant learnings and decisions, including status and confidence when available, followed by one or two sentences of synthesis. If nothing relevant exists, say so plainly.