Skip to main content

Mind Map & Data Relationship Index

The mindmap is Reggie’s lookup atlas.
It describes where every fact in RABS lives, what shape it takes, and how different domains connect.
Rather than storing raw data, the mind-map stores metadata, allowing any module—or the LLM—to jump directly to the right table, context bucket, or API with a single semantic query. The mindmap is **Reggie’s lookup atlas**. It describes **where** every fact in RABS lives, **what** shape it takes, and **how** different domains connect. Rather than storing raw data, the mindmap stores *metadata*, allowing any module—or the LLM—to jump directly to the right table, context bucket, or API with a single semantic query.

1. Purpose​

NeedHow mindmap helps
Fast RAG retrievalAvoids brute-force scans; the agent can resolve “client risk notes” directly to the clients.notes column.
Cross-domain joinsEncodes FK and semantic links so analytics tooling can auto-generate JOIN clauses.
Self-documentationLiving diagram of the data estate; new devs see the layout in one file.
Change-impact analysisWhen a table or field changes, the mind-map lists every dependent module and prompt template, preventing breaking changes.

| **Self-Documentation** | Provides a living diagram of the entire data estate; new developers can see the system layout in one file. |

| **Change Impact Analysis** | When a table or field changes, the mindmap shows every dependent module and prompt template, preventing breaking changes. |

2. Storage & Format​

JSON Shape (excerpt)​

"tables": { "internalmonologue": { "desc": "System-wide event stream", "vectorCol": "vector", "links": [ { "to": "reasonlog", "type": "causal", "via": "event_reference" } ] } }, "buckets": { "rostering_lessons": { "module": "rostering", "desc": "Context bucket of past roster-allocation mistakes." } }, "external": { "twilio.sms": { "type": "REST", "desc": "Outbound / inbound SMS logs" } }   "links": [

  { "to": "reasonlog", "type": "causal", "via": "event_reference" }

3. How It’s Used​

  }

  • LLM function calls – getDataPath(entity:string) resolves natural-language queries like “booking history” to a concrete path tables.bookings.
  • Code generation – Dev scripts read the mind-map to create TypeScript types or stub DB migrations.
  • Vector retrieval – The RAG helper lists all objects with a vectorCol then performs unified similarity search across them.
  • Data lineage – When a user asks “why was this value calculated?”, the mind-map provides the dependency graph.   },

4. Maintenance​

  },

  "external": {

  • scripts/update-mindmap.ts --table new_table – run after a DB migration.
  • scripts/sync-openapi.mjs – ingests an OpenAPI spec (e.g., Twilio) to map external endpoints.
  • CI job regenerates an SVG graph of the map and commits it to the repository.
---

## 🔗 Related Docs

* [04_Context_Buckets_Lessons_Learned.md](./04-context-buckets-lessons)  
* ../03_LLM_&_Prompts/01_Central_LLM_Gateway.md



    Code Generation: Development scripts can read the mindmap to automatically create TypeScript types or stub out database migration files.



    Vector Retrieval: The RAG (Retrieval-Augmented Generation) helper queries the mindmap for all objects with a vectorCol, then performs a unified similarity search across multiple tables.



    Data Lineage: When a user asks "why was this value calculated?", the mindmap provides the dependency graph to trace the data's origin.



4\. Maintenance



The mind-map is kept up-to-date via automated scripts:



    scripts/update-mindmap.ts --table new\_table is run after a database migration.



    scripts/sync-openapi.mjs can ingest an OpenAPI spec from an external service (like Twilio) to map its endpoints.



    A CI job can regenerate a visual SVG graph of the map and commit it to the repository.



🔗 Related Docs



    04\_Context\_Buckets\_Lessons\_Learned.md



    ../03\_LLM\_and\_Prompts/01\_Central\_LLM\_Gateway.md