You need a local Codex client and an Agiler API key restricted to your intended project. Use projects:read for the first connection.
Set the key and register Agiler
Make AGILER_API_KEY available to the Codex process. In a POSIX shell:
export AGILER_API_KEY='ak_REPLACE_WITH_YOUR_KEY'
codex mcp add agiler --url https://mcp.agiler.io/v1 --bearer-token-env-var AGILER_API_KEY
Replace the key placeholder locally. Use your normal secret manager or private environment configuration for ongoing use; a shell may save the export command in history.
The equivalent entry in your personal ~/.codex/config.toml is:
[mcp_servers.agiler]
url = "https://mcp.agiler.io/v1"
bearer_token_env_var = "AGILER_API_KEY"
Use either registration method. Merge the entry with existing configuration and avoid duplicate mcp_servers.agiler tables. Codex sends the environment variable’s value as a bearer token. This connection does not need codex mcp login.
Verify the connection
Start a new Codex session in the same environment. Run codex mcp list or use /mcp in the CLI, then ask:
Use Agiler’s whoami tool to check my account, then list the projects I can access. Do not make changes.
Continue with your first task.
Use the IDE extension
Local Codex clients on the same host share the configuration. The IDE process also needs AGILER_API_KEY in its environment; a variable exported in a separate terminal may not reach an already-running editor. Restart the editor with the key available.
For project-scoped settings, use .codex/config.toml in a trusted project and keep the key in the environment. Cloud environments need their own supported configuration and credentials; a local setting alone does not configure cloud access.
See OpenAI’s MCP documentation for client configuration. Add project instructions for recurring work, or use troubleshooting if tools are missing.