Since updating to 1.227.0, every Codex agent (existing, resumed and newly spawned) exits 0 s after launch: "'codex' exited 0s after launch [ExternalScript] ...\codex.ps1 (exit code 1)". Running codex with the project's AgentsRoom CODEX_HOME reproduces it: "failed to load bootstrap configuration / Caused by: url is not supported for stdio / in mcp_servers.AgentsRoom-MCP". The generated config declares AgentsRoom-MCP with a url plus bearer_token_env_var; the merged config apparently also carries a stdio command for the same server. Codex CLI 0.163.0 works normally with its default home. Claude agents are unaffected. Expected: Codex agents start as before 1.227.0.
AgentsRoom diagnostics
- App version: 1.227.0
- OS: win32 10.0.26200 (x64)
- Agent CLI: claude
- App language: system
Expected behavior
Codex agents start and resume normally, as before 1.227.0.
Actual behavior
Every Codex agent exits immediately with exit code 1; status shows failed.
Steps to reproduce
- Update AgentsRoom to 1.227.0 and restart. 2. Resume or spawn any Codex agent. 3. It exits 0 s after launch. 4. Set CODEX_HOME to the project's AgentsRoom codex folder and run 'codex mcp list': same bootstrap error.
Version or build
AgentsRoom 1.227.0, Codex CLI 0.163.0, Windows 11
Errors and logs
Error: failed to load bootstrap configuration
Caused by:
url is not supported for stdio
in mcp_servers.AgentsRoom-MCP