mcp_engram_thought_tile_create
Create a new Thought Tile to store structured knowledge with bonds and tensor mirror for agent recall, momentum, and auto-linking to primary intent. Supports multiple tile types.
Instructions
Create a new Thought Tile (textual functor payload optimized for agent recall, momentum, NREM, and ki_hijacker). Dual-writes a tensor:tile__ mirror with bonds (goal/trace/spatial). Supports research_offload, state_machine, tabular, knowledge_graph, formal_spec, propose_improvement (verified tensor update on target). Pair with thought_tile_create_visualization for rich human-viewable companion. Auto-links to Primary Intent.
Input Schema
TableJSON Schema
| Name | Required | Description | Default |
|---|---|---|---|
| title | Yes | Short human-readable title for the tile | |
| payload | Yes | Structured JSON payload matching the schema for the chosen tile_type | |
| tile_type | Yes | research_offload | state_machine | tabular | knowledge_graph | formal_spec | propose_improvement | html_visualization | verified_sequence | agent_response (RPT v3 — use mcp_engram_turn_record for turn envelope) | |
| parent_tile | No | Optional parent tile for decomposition / result hierarchy | |
| goal_context | No | Optional explicit goal. If omitted, auto-links using primary_goal + recent active goal logic (same as record_reasoning_trace). | |
| process_context | No | Optional process:engram.* key — emits realized_by edge (WS-3) | |
| spatial_references | No | Optional list of existing concept names (spatial AST nodes, ritual anchors, etc.) this Tile compresses or references. Creates compresses_path / compresses_chain_from relations for trace:* refs. |