- Home
- Skills
- Agents & Orchestration
- Cortex Deliberation: Multi-Agent Tech & Code Reviewer
Works with the AI Agents you already use
Cortex Deliberation: Multi-Agent Tech & Code Reviewer
Stop AI hallucinations in technical decisions. A multi-agent skill that simulates a Visionary, an Auditor, and a Validator to deliver evidence-based code reviews and architecture choices.
Free
Cortex Deliberation: Multi-Agent Tech & Code Reviewer
Example session with this skill installed
Analyze our plan to migrate from a monolithic Postgres instance to a sharded Vitess setup for our multi-tenant SaaS. Focus on data consistency and migration risks.
- Read your context and instructions
- Compiled the cortex deliberation
I have simulated a deliberation between the Visionary and Auditor personas regarding your Vitess migration strategy.
Technical Deliberation Report: Vitess Migration
1. Visionary Proposal
The proposed migration utilizes Vitess to achieve horizontal scaling by sharding based on tenant_id. This approach maintains SQL compatibility while offloading connection management to VTGate.
2. Auditor Critique
| Finding | Evidence | Impact | Severity | Confidence |
|---|---|---|---|---|
| Cross-shard Join Latency | Lack of explicit lookup vindexes | Query timeouts | High | High |
| Distributed Transaction Overhead | 2PC requirements for global reports | Performance degradation | Medium | Medium |
| Sequence Collision | Reliance on auto-increment IDs | Primary key conflicts | Critical | High |
3. Synthesis & Recommendation
The migration is technically sound but requires immediate changes to the schema design before execution. The primary risk is the current reliance on native Postgres sequences which will collide in a sharded environment.
Recommendation: Proceed with migration only after implementing Vitess Sequences (vindexes) for all primary keys.
Implementation Plan:
- Refactor Identifiers: Implement 64-bit BigInt sequences or UUIDs to ensure global uniqueness.
- Vindex Selection: Define functional vindexes for
tenant_idto ensure data locality. - Validation: Run
VReplicationin dry-run mode to measure replication lag.
Next steps
- Update the schema definition to replace serial types with Vitess-compatible sequences.
- Map all cross-tenant queries to identify where
scatterqueries can be refactored into targeted lookups. - Define the
MoveTablesworkflow parameters for the initial data sync.
Connects securely to your tools. The creator never sees your data.
About this skill
Stop letting AI give you rushed, one-sided answers. Cortex Deliberation brings your technical decisions to a professional standard by simulating a rigorous multi-agent debate.
Based on the Cortex framework, this skill forces the AI to adopt distinct roles:
- The Visionary: Proposes scalable, modern solutions.
- The Auditor: Attacks the proposal using a strict security checklist (Auth, Secrets, Input Validation) and demands evidence for every risk.
- Technical Validation: Cross-checks claims against official documentation to eliminate AI hallucinations.
- The Synthesizer: Merges the debate into a final, evidence-based decision.
Key Features:
- Includes a Fact vs. Hypothesis separation phase.
- Requires evidence and severity ratings for security flaws.
- Generates an executable plan with Acceptance Criteria, Testing, and Rollback strategies.
- Adapts its output: provides a full report for complex architectures, or a quick answer for simple questions.
Perfect for use in Cursor, Claude Code, and Copilot when designing architectures or doing critical code reviews.
Find more skills like this on the Claude Skills Marketplace, the Codex Skills Marketplace and the Gemini Skills Marketplace, or browse Claude Skills for Agents & Orchestration.
By default, operates in Simulated Mode (one model role-playing all agents). Independent multi-agent execution requires an external orchestrator. Web searches depend entirely on the host agent's capabilities.
How to use
- 1
Connect your agent to the Agensi MCP
Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup
- 2
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 3
Ask for what you need
Your agent loads this skill and uses it on your task. It always asks before buying a skill.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill goes through the trust checks below.
Security scanned
Checked by our 8-point scan before listing
Fresh listing
Recently published to Agensi
Free to use
No account required to browse
Trust & safety
Security scan
Security scan passed 19 days ago
- Free with an Agensi account
Permissions
The skill operates strictly on the context provided in the chat. It does not execute arbitrary code, require API keys, or access external networks without host agent permission.