Traction scoreGitHub stars can be faked, so popularity alone can be misleading. Traction Score looks for broader signs of recent attention, adoption, and active maintenance.
This collection provides specialized tools for developers integrating external agents with the Codex App Server. It streamlines the technical overhead of maintaining model health and managing long-running task contexts.
By providing dedicated mechanisms for installation audits and context rollovers, the skills help automation builders ensure that headless calls from agents remain stable and efficient.
WHO IT'S FOR
AI agent / automation builders
integrate external agents with Codex App Server
Platform / DevEx teams
audit Codex installation health and context quality
Compatible AgentsThe repository documents support for these agents. The skills may also work with other agents that can load SKILL.md files, but they may need some setup or small changes.
Installs: Only the codex-cli-model-bridge skill, installed globally for Codex. Its payload also supports Claude Code and generic Agents-compatible harnesses, but the supplied evidence does not document separate installation procedures for those harnesses. · Codex
Before you start
Codex CLI
Python 3.11 or newer
A loopback CLIProxyAPI, or an existing Codex Router, already exposing the target models
Preserved ChatGPT login when Codex Desktop history should remain on the openai provider
Platform: Windows is supported through the isolated cli-proxy profile; Homebrew, LaunchAgents, and Codex Router are optional.
Run this complete multiline command in a terminal.
Installs: Only the codex-doctor skill, installed globally for Codex. · Codex
Before you start
Codex CLI with codex doctor --json
Python 3.11 or newer
Git
Platform: macOS or Linux
Run this command in a terminal.
npx skills add zjp1997720/zhijian-skills -g -a codex --skill codex-doctor -y
After installation, invoke $codex-doctor in a Codex session or ask Codex to run a workspace health check. Do not type /doctor because the skill is not registered as a slash command.
A bounded task description replacing [describe the task]
At least two potentially independent deliverables
Acceptance criteria for each deliverable
Any applicable permission, model, provider, or live child-slot constraints
Use $codex-model-routing-team for this bounded task: [describe the task]. Identify two independent, read-only deliverables with clear acceptance criteria, determine whether delegation has positive net benefit, and either compile and execute the validated team plan or explain why direct execution is the correct route. Keep final integration and acceptance with the main agent.
Turn dense material into a self-contained comparison report
Uses html-express
Source material or a readable file path replacing [source material or file path]
The alternatives or dimensions to compare
An output filename replacing [output file]
Check the repository for this skill’s setup.
Use $html-express to turn [source material or file path] into one self-contained HTML comparison report. Use a comparison table plus only the other components needed for readability, preserve factual uncertainty as [数据待补：说明], and save the result as [output file].html.
A complete visual prompt replacing [complete visual prompt]
An output directory replacing [output directory]
Codex CLI already installed and authenticated with codex auth login
A readable ~/.codex/auth.json
Network access to the documented Codex Responses and OAuth endpoints
Check the repository for this skill’s setup.
Use $codex-image-gen to generate a landscape illustration from this prompt: [complete visual prompt]. Use high quality, save it under [output directory], and report the resulting image path and whether any input or reference image was used.
Claude
Codex
Gemini CLI
OpenCode
Grok
DeepSeek
ZCode