Codex

Tested with Codex CLI 0.163: sign-in, installed server, hosted server with a key, skills, and Codex running on a Helipod AI Gateway model calling Helipod tools.

RequirementsCopy Anchor Link

• Codex (CLI or IDE extension; both read ~/.codex/config.toml).
• Node.js 18 or newer for the local server.
• HELIPOD_API_KEY in your environment, see Get a key.

Add the serverCopy Anchor Link

[mcp_servers.helipod]
url = "https://api.helipod.io/mcp"

codex mcp login helipod

Add the skillsCopy Anchor Link

npx skills add helipodapp/mcp -g -a codex

This puts them in ~/.agents/skills, which Codex reads. Use it without -g inside a repository to add them to that repository only (.agents/skills).

Check it worksCopy Anchor Link

• codex mcp list shows helipod as enabled; codex mcp get helipod shows its settings.
• In Codex, /mcp lists the Helipod tools.
• Ask: "List my Helipod projects."

Approvals and sandboxCopy Anchor Link

Codex asks before MCP tool calls depending on your approval mode. Keep approvals on (--ask-for-approval on-request or the default) when the key can spend money. --dangerously-bypass-approvals-and-sandbox runs every tool without asking; do not use it with a Helipod key that can create or delete resources.

Run Codex on Helipod AI modelsCopy Anchor Link

Codex can use models from the Helipod AI Gateway (through its Responses API), billed per token in Rupiah. Add the provider:

[model_providers.helipod]
name = "Helipod AI Gateway"
base_url = "https://ai.helipod.io/v1"
env_key = "HELIPOD_AI_KEY"
wire_api = "responses"

and run codex -c model_provider=helipod -m glm-5.3, or set model = "glm-5.3" and model_provider = "helipod" at the top of the file. HELIPOD_AI_KEY is an AI Gateway key (hel_ai_…), not the hel_ API key. See Coding tools → Codex.

TroubleshootingCopy Anchor Link