AI Agents That Work With OpenAI Codex (2026)
As of Aug 10, 2026, 129 AI agents in the KanonAgent index are judged to connect to OpenAI Codex; this page covers the top 11 by real traction (11 with the source line quoted). Judgements come from quoted evidence, not vendor claims — where we cannot read it, we leave it blank.
Connecting agents to OpenAI Codex lets developers run it in parallel with other models, delegate code review or generation tasks, and embed it inside IDEs or orchestration layers. In 2026 the pattern is shifting from single-model use to fleets where Codex handles specific subtasks while other agents manage context, diagrams, or deployment.
Updated 2026-08-10 · 11 products · live data from KanonAgent
1. orca41.2k upvotes
Orca runs Codex side-by-side with Claude Code and others so you can orchestrate parallel coding agents across desktop and mobile.
source: “Run Codex, ClaudeCode, OpenCode or Pi side-by-side”
2. hallmark19.4k upvotes
Hallmark applies its anti-slop design system to Codex output, raising readability and quality of generated code.
source: “Claude Code, Cursor, and Codex”
3. ai-berkshire15.3k upvotes
AI Berkshire feeds investment research prompts to Codex (and Claude Code) for multi-agent adversarial analysis.
source: “built for Claude Code / Codex”
4. html-anything8.2k upvotes
html-anything auto-detects Codex on your PATH so your local agent can generate and ship HTML pages without manual coding.
source: “9 coding-agent CLIs auto-detected on your PATH (Claude Code · Cursor Agent · Codex · Gemini CLI · GitHub Copilot CLI · OpenCode · Qwen Coder · Aider ·”
5. archify7.7k upvotes
Archify lets Codex CLI generate architecture diagrams with theme toggles and export options for quick system visualization.
source: “Raven, Cursor, Claude Code, Codex CLI, and OpenCode”
6. cursor-talk-to-figma-mcp7.0k upvotes
cursor-talk-to-figma-mcp gives Codex direct MCP access to read and edit Figma files, cutting manual design handoff work.
source: “MCP integration between AI Agent (Cursor, Claude Code, Codex) and Figma”
7. marketingskills6.4k upvotes
marketingskills trains Codex on CRO, copywriting and SEO so the agent can execute growth-engineering tasks.
source: “Works with Claude Code, OpenAI Codex, Cursor, Windsurf”
8. openwork5.2k upvotes
openwork adds an MCP layer to Codex for team-based workflow orchestration and multi-role task handoff.
source: “Add one OpenWork MCP to Codex, Claude Code, Cursor”
9. codex-plugin-cc4.7k upvotes
codex-plugin-cc lets you call Codex from inside Claude Code to review code or delegate tasks without leaving the editor.
source: “Use Codex from inside Claude Code”
10. ego-lite2.7k upvotes
ego-lite runs Codex inside a browser built for parallel human-AI work, keeping context and execution in one window.
source: “with your AI agents, like Codex or Claude Code”
11. agents-cli2.3k upvotes
agents-cli turns Codex into an expert at building and deploying agents on Google Cloud with one CLI.
source: “Works seamlessly with: Antigravity CLI • Claude Code • Codex”
Before you wire it up
Choose by how deeply the agent calls Codex (CLI detection, MCP, or plugin) versus treating it as one interchangeable model. Check permission scope: agents that need file-system or browser access require broader auth than pure code-review tools. Watch rate limits on Codex calls when running multiple agents in parallel; most failures come from context-window exhaustion or lost state between Codex and the orchestrator.
FAQ
Do I need separate API keys for each agent that uses Codex?
Most agents reuse your existing OpenAI key; only those with MCP or plugin layers add extra auth steps.
What breaks most often when wiring agents to Codex?
Context loss between agents and Codex hitting token limits during long multi-file tasks.
Can Codex be called from Claude Code without leaving the editor?
Yes, via the codex-plugin-cc integration.
How is this list ranked?
By real traction from our index (upvotes / MRR / growth rate, whichever the product actually has) — not editorial picks, and we do not accept paid placement. Sources: ProductHunt, Hacker News, GitHub, HuggingFace, Reddit, TrustMRR.
What is the inclusion bar?
A page is published only when at least 5 real products qualify. Judgement fields (autonomy, prerequisites, cost, integrations) all require a source quote — where we cannot read it, we leave it blank rather than guess.
How often is this updated?
Collection runs continuously; this page was regenerated on 2026-08-10. Every number is verifiable through our public read-only API: https://kanonagent.com/data