| name | graphify-memory |
| description | Read and write to the shared cross-agent memory graph using graphify. |
Graphify Memory Skill
Teaches the agent to read and write to the unified cross-agent memory graph (Token-Agnostic Handoff Protocol). This ensures that context persists between sessions and across different AI models (Antigravity, Claude Code, OpenCode).
When to use this skill
- When starting a session in a project that uses shared memory.
- When you are about to run out of context or end a session, you must write a summary of what you did.
- Before making significant architectural decisions, to see what was previously decided.
How to use Graphify
The project root must contain graphify-out/graph.json or you should be operating in a workspace that supports the graphify CLI.
Querying Memory
To query historical context:
graphify query "topic" --graph graphify-out/graph.json
To find paths between concepts:
graphify path "concept A" "concept B" --graph graphify-out/graph.json
To explain a concept:
graphify explain "concept" --graph graphify-out/graph.json
Writing to Memory
Always write your findings, completed tasks, and architectural decisions to the graph. Do not let context die in the session.
Unified Memory Rules
- Context window is ephemeral; disk is eternal.
- If you learn a new pattern, add it to the graph.
- If you fix a tricky bug, document the fix in the graph.