Xpert — Development skill for Claude Code
Packx + Codex expert agents for code analysis.
How to install Xpert
This entry records only its repository, not the path inside it, so there is no
exact command to give. Open johnlindquist/xpert and copy the folder into
~/.claude/skills/, or the file into ~/.claude/agents/.
What Xpert does
Packx + Codex expert agents for code analysis.
Alternatives in Development
- Scientific Agent Skills — A set of ready to use Agent Skills for research, science, engineering, analysis, finance and writing 18.7k ★
- Income Investment — Income Investment: Durable and Opportunistic Distribution Analysis 15.9k ★
- Cohort Analysis — Cohort retention curves, feature adoption, and segment insights 7.8k ★
README
xpert (codex-packx-expert)
A Bun/TypeScript CLI that orchestrates **packx → Codex expert agents** to produce:
- An expert-friendly, copy/paste-ready markdown bundle in
.xpert/bundles/.md - A single expert instruction file in
.xpert/plans/.md
Designed for use from tools like Claude Code (clean stdout, progress on stderr).
Requirements
buninstalledcodexinstalled + authenticated (codex login)packxinstalled (used by the bundle stage)
Install
bun install
bun link # Installs globally as `xpert`
Commands
All commands available as `xpert ` or `xpert-`:
Doctor
xpert doctor
Checks for `codex`, `packx`, and authentication.
Bundle only
xpert bundle "Review the notes window focus logic"
# or stdin
cat prompt.md | xpert bundle -
Output:
OUTPUT_FILE_PATH: .xpert/bundles/.md
Add `--dump` to print the whole bundle to stdout for copy/paste.
Expert only (from an existing bundle)
xpert expert --bundle .xpert/bundles/.md ""
End-to-end (bundle → expert)
xpert solve "Review the notes window focus logic"
Output:
OUTPUT_FILE_PATH: .xpert/plans/.md
Add `--dump` to print the full expert instructions markdown.
Options
--cd Workspace root (default: .)
--name Output name (default: derived from request)
--model Codex model override
--dump Print generated markdown to stdout
--dump-only Print markdown only (no OUTPUT_FILE_PATH line)
--raw-limit packx token cap (default: 47000)
--concurrency Scout concurrency (default: 2)
--iterations Audit/revision passes (default: 2)
--no-search Disable web search for docs-researcher
Safety defaults
- Workspace-write sandbox: Codex can write to the workspace (required for bundle output)
- AGENTS.md disabled: Local project
Related Skills
Siftlight
Context-efficient project search and bounded code analysis for Claude Code & Codex & Pi & OMP & kimi ....and r
Baoer Signal Grep
Context-efficient project search and bounded code analysis for Claude Code & Codex & Pi & OMP. and remote MCP
Decant
Local-first analysis of Claude Code and Codex sessions: token spend, context windows, files touched, and cost.
Targets R Skill
Claude Code and Codex skill for the targets R package (targets = 1.12). Modern patterns for reproducible analy
MARE MCP Toolbox
Agentic malware analysis environment with MCP-connected disassemblers, RE tooling, and structured workflows fo
Agents Analysis
Effective token price reports for Codex and Claude agent subscriptions
Related Agents
Agentic Malware Analysis
Agentic malware analysis environment with MCP-connected disassemblers, RE tooling, and structured workflows fo
Codex Deep Reasoner
CLI-backed read-only deep-reasoning worker — the codex variant of deep-reasoner. Use for root-cause analysis,
Codex Analyst
Token-heavy READ-ONLY analysis, codebase surveys, audits and inventories, executed by the Codex CLI rather tha