Oracle Research Skill

This skill enables confidential research for Oracle consulting work, with built-in OCI knowledge and abstraction protocols.

Research Protocols

Quick Research (5 min)

Single WebSearch query
Summarize top 3 findings
Note one actionable insight

Standard Research (15 min)

3-5 WebSearch queries with variations
Cross-reference findings
Create topic file in research/topics/
Link to projects if relevant

Deep Dive (30+ min)

Comprehensive search strategy
5+ sources analyzed
Synthesis with patterns identified
Recommendations documented
Follow-up research noted

Search Strategy by Domain

OCI Services

"OCI [service] best practices 2026"
"Oracle Cloud [service] architecture patterns"
"[service] vs [competitor] enterprise comparison"

Generative AI

"enterprise GenAI architecture 2026"
"LLM [use case] production deployment"
"RAG implementation [industry] enterprise"

Kubernetes/Cloud Native

"OKE [feature] production patterns"
"Kubernetes [pattern] enterprise scale"
"service mesh [option] OCI integration"

Industry-Specific

"[industry] AI use cases 2026"
"[industry] cloud transformation patterns"
"Oracle [industry] solutions"

OCI Service Quick Reference

Compute

Service	Use Case	Key Feature
Compute VMs	General workloads	Flexible shapes
Bare Metal	High performance	Dedicated hardware
GPU Instances	AI/ML training	A100, A10, L40S
Container Instances	Serverless containers	No cluster overhead

AI/ML

Service	Use Case	Key Feature
OCI Generative AI	LLM inference	Cohere, Meta models
AI Services	Pre-built AI	Vision, Language, Speech
Data Science	ML development	Notebooks, jobs, models
Dedicated AI Clusters	Large-scale AI	Reserved GPU capacity

Data

Service	Use Case	Key Feature
Autonomous DB	Self-managing	Auto-tuning, patching
MySQL HeatWave	Analytics	In-memory acceleration
NoSQL	Key-value/JSON	Massive scale, low latency
Object Storage	Data lake	S3-compatible, tiers

Containers

Service	Use Case	Key Feature
OKE	Kubernetes	Managed control plane
Virtual Nodes	Serverless K8s	No node management
Container Registry	Image storage	Scanning, signing

Confidentiality Quick Rules

NEVER in Research

Customer names in search queries
Customer-specific configurations
Internal Oracle pricing
Unreleased product details

ALWAYS Abstract

Real	Abstracted
Customer name	Codename (K, V, E) or "the customer"
Specific amount	Range or "significant"
Exact date	Quarter (Q1, Q2)
Specific region	"primary region"

Safe to Include

Public OCI documentation
Oracle blog posts
Conference talks (public)
Generic architecture patterns
Industry trends (non-customer specific)

Research Output Templates

Topic Research

# [Topic]

**Date:** YYYY-MM-DD
**Type:** Quick / Standard / Deep Dive
**Project:** K / V / E / General

## Summary
[2-3 sentences]

## Key Findings
1. [Finding with source]
2. [Finding with source]

## Application
[How this applies to current work]

## Sources
- [URL 1]
- [URL 2]

Competitive Research

# [Topic]: OCI vs Alternatives

**Date:** YYYY-MM-DD
**Context:** [Why comparing]

## Comparison Matrix
| Criteria | OCI | Alternative |
|----------|-----|-------------|
| [Criterion] | [OCI position] | [Alt position] |

## OCI Advantages
- [Advantage 1]

## Areas to Address
- [Gap 1]

## Positioning Guidance
[How to discuss with customers]

Mandatory GitHub Resources

ALWAYS check these before recommending custom solutions:

Repository	URL	When to Check
OCI AI Blueprints	https://github.com/oracle-quickstart/oci-ai-blueprints	GenAI on OKE, vLLM, Llama Stack, fine-tuning
AI Accelerator Packs	docs/ai_accelerator_packs	Route optimization, video search, enterprise reasoning
technology-engineering	https://github.com/oracle-devrel/technology-engineering	Solution patterns, architecture guides
oracle-quickstart	https://github.com/oracle-quickstart	Terraform modules, reference architectures

Customer GenAI Decision Flow

Customer needs GenAI deployment?
├── Standard LLM inference → OCI AI Blueprints (vLLM)
├── Complete agentic stack → OCI AI Blueprints (Llama Stack)
├── Fine-tuning → OCI AI Blueprints (LoRA Fine-Tuning)
├── Route optimization → AI Accelerator Pack (cuOPT)
├── Video analysis → AI Accelerator Pack (VSS)
├── Enterprise chat agent → AI Accelerator Pack (AI-Q)
└── Unique requirements → Custom with oracle-devrel patterns

Integration Points

Link to Projects

When research relates to K, V, or E:

Save in research/projects/[CODENAME]/
Update project file's Notes section
Suggest application abstractly

Feed Synthesis

Weekly, review research/topics/ and create: research/synthesis/week-YYYY-WW.md

Support Daily Capture

Research findings can become:

Learnings in daily capture
Skills developed in project files
Content for reports

This skill is automatically loaded when research context is detected.

Cline Activation

To use this skill in Cline, reference it at the start of your message:

@skills/oracle-research/SKILL.md

Research OCI Generative AI pricing options for a high-volume document processing use case.

Or in a .clinerules workflow:

## Research Tasks
When asked to research Oracle or OCI topics, load @skills/oracle-research/SKILL.md and follow the Standard Research protocol. Save output to research/topics/.

Triggers: Oracle research, OCI comparison, architecture patterns, GenAI use cases