claude-phd-panel

Claude PhD Panel Plugin

Academic expert perspectives for any codebase. Seven specialized review commands that analyze your project from different research domain viewpoints.

Commands

Command	Role	What it reviews
/cs	Computer Science PhD	Algorithm complexity, data structures, concurrency correctness
/db	Database Theory PhD	Schema normalization, query optimization, consistency models
/stats	Statistics PhD	A/B test design, metric validity, statistical significance
/ds	Data Science PhD	ML pipelines, feature engineering, model evaluation
/dist-sys	Distributed Systems PhD	Consensus, fault tolerance, partition handling
/pl	PL Theory PhD	Type safety, abstraction design, error handling patterns
/sec	Security & Cryptography PhD	Primitive selection, protocol safety, key management, side channels

Utility commands

Command	Purpose
/config	Manage ~/.claude/claude-phd-panel.json — show, init, or read a setting
/audit	Verify all command files conform to plugin conventions

Installation

/plugin marketplace add JFK/claude-phd-panel-plugin
/plugin install claude-phd-panel

Usage

Review mode

Run any command to get a full academic review:

/claude-phd-panel:cs                   # Full CS review of current repo
/claude-phd-panel:cs owner/repo        # Analyze a specific repo
/claude-phd-panel:db schema            # Focus on schema design
/claude-phd-panel:stats metrics        # Focus on metrics/experiments

Question mode

Ask any PhD a direct question — they'll answer from their academic perspective, grounded in your actual codebase:

/claude-phd-panel:cs Is this sorting approach optimal for our data size?
/claude-phd-panel:db Should we denormalize this table for read performance?
/claude-phd-panel:stats Is our A/B test sample size sufficient?
/claude-phd-panel:ds Is there data leakage in our feature pipeline?
/claude-phd-panel:dist-sys Can this service handle network partitions?
/claude-phd-panel:pl Should we use generics here or concrete types?
/claude-phd-panel:sec Is HS256 safe for our JWT use case?

Language configuration

Configure the output language for all PhD commands:

/claude-phd-panel:config init    # Create ~/.claude/claude-phd-panel.json

Edit language to your preferred ISO 639-1 code (ja, zh, ko, etc.):

{
  "language": "ja"
}

All review output and Q&A responses will be in your chosen language. Code blocks, file paths, and structural headings stay in English for tool compatibility.

Cross-referencing with C-Suite

Run PhD Panel commands alongside Claude C-Suite commands in the same session for executive + academic perspectives:

/claude-c-suite:cto                    # CTO flags DB performance concern
/claude-phd-panel:db                   # DB PhD validates with academic rigor

Worked example: validating a CTO concern with academic rigor

A typical session combining both plugins:

1. CTO flags an issue — /claude-c-suite:cto reports that the user profile endpoint is slow under load and recommends caching:

   "Profile reads dominate p95 latency. Consider Redis caching with a 5-minute TTL. Severity: High."

2. DB PhD validates the diagnosis — /claude-phd-panel:db Is caching the right fix here, or is the underlying query the problem? reads the actual schema and queries, then responds:

   "The users table is missing an index on (org_id, last_active_at), which forces a sequential scan on every profile read. Caching would mask the problem but not fix it — and would introduce staleness for a workload that has strong read-your-writes expectations. Recommend adding the composite index first; reassess caching only if p95 is still high."

3. CS PhD checks the algorithmic side — /claude-phd-panel:cs notices the handler also does an O(n²) merge of friend lists in application code and flags it as a separate, larger issue.

4. CEO synthesizes — /claude-c-suite:ceo combines all three findings: index first (cheap, correct), then refactor the merge (medium effort, bigger win), defer caching (don't mask root causes).

The pattern: C-Suite identifies what to worry about; PhD Panel proves whether the proposed fix is actually the right one.

Related plugins

• claude-c-suite-plugin — Executive perspectives (CEO/CTO/CSO/CFO/CMO/CLO/etc.) for code review. C-Suite commands automatically incorporate PhD Panel findings when run in the same session.
• expert-craft-plugin — Create your own custom domain experts via interactive dialogue. Pairs naturally with PhD Panel and C-Suite for project-specific reviews.

Design Principles