Add oracle-data-studio-mcp-server (Essbase + ADP + Data Transforms)

[tsikinov]

Proposal

Contribute a new Python MCP server covering three Oracle Data Studio
services with 60 high-level composite tools.

Capabilities

• Essbase — 30 tools: apps, cubes, outline, MDX, calc, security, jobs
• ADP (Autonomous Database Data Platform) — 15 tools: Analytic Views,
  Select AI, Insights, cloud loading, catalogs, sharing, Oracle 23ai
  annotations
• Data Transforms — 15 tools: pipelines, schedules, connections,
  workflows, dataloads
• 1 prompt template (adp_sql_with_annotations) for annotation-aware
  SQL generation

Differentiator: annotation-driven query routing

Ships server-level instructions that nudge LLM clients (Claude
Desktop / Cursor / Codex) to consult Oracle 23ai annotations before
generating SQL — fixing common silent bugs (units, aggregations, join
keys, time grain, PII).

A second pattern (cube=… / analytic_view=… annotations on fact
tables) lets the LLM deterministically route aggregate questions to
the right query source — Essbase cube vs. Analytic View vs. raw
table — without name-matching or guessing.

Validation

A/B tested live on Oracle Autonomous Database 23.26 with the MovieLens
dataset and an Essbase cube of the same name — annotation-aware SQL
generation beat naive SQL generation 5/5 across realistic failure
modes (unit mismatch, grain mismatch, aggregate-function choice, join
key, PII avoidance).

109 unit tests, all passing. Independent security/bug audit: all
High-severity findings resolved before submission.

Branch

https://github.com/tsikinov/mcp/tree/add-data-studio-mcp-server

Will follow with a PR once this issue gets a number.