Enable Codex update_plan and cap MCP output_token_limit
Codex CLI rust-v0.152.0 makes update_plan opt-in, adds per-tool MCP output_token_limit with resume-safe truncation, allows package-style MCP names (: @ / .), and supports thread/shellCommand timeouts over one hour. Update team config so planning, MCP, and long shell jobs keep working.Why this loop
OpenAI Codex CLI rust-v0.152.0 turns the planning tool off unless tools.update_plan.enabled = true; teams that relied on plan updates will lose them without that flag. Individual MCP tools now accept output_token_limit with truncation that stays consistent across session resumes—set it on verbose servers so context does not balloon. MCP server names may contain :, @, /, and ., so package-style names work in CLI commands and authentication; rename servers that were previously forced into alphanumerics. App-server clients can configure thread/shellCommand timeouts, including deadlines longer than one hour, which unblocks long builds. Approval-heavy workflows should resume existing threads: automatic approval reviews retain longer messages and a larger transcript, and they preserve user instructions, answers, and valid authorizations across history compaction, while resumed threads restore the saved working directory.
Proposed actions
- In Codex config, set tools.update_plan.enabled = true so the planning tool stays available; rust-v0.152.0 disabled it by default.
- On each MCP tool that returns large payloads, set output_token_limit so truncation stays consistent when sessions resume.
- Register MCP servers with package-style names that include :, @, /, or . (for example @scope/pkg) in CLI commands and authentication; those characters are now valid.
- For app-server clients, set thread/shellCommand timeouts above one hour for any job that must run longer than the previous deadline.
- Resume existing threads instead of starting new ones so saved cwd, MCP tool caches, and Guardian approvals (instructions, answers, authorizations) survive compaction.
Agent prompt
Paste into your agent or query via MCP (get_agent_prompt) — free, no extra AI costCodex / agent CLI task
DevAgentRadar → OpenAI Codex
Task: evaluate and optionally implement follow-ups from this coding-assistant release.
Context
Assistant: OpenAI Codex CLI Proposal: Enable Codex update_plan and cap MCP output_token_limit Summary: Codex CLI rust-v0.152.0 makes update_plan opt-in, adds per-tool MCP output_token_limit with resume-safe truncation, allows package-style MCP names (: @ / .), and supports thread/shellCommand timeouts over one hour. Update team config so planning, MCP, and long shell jobs keep working. Primary source: https://github.com/openai/codex/releases/tag/rust-v0.152.0
Why it matters
OpenAI Codex CLI rust-v0.152.0 turns the planning tool off unless tools.update_plan.enabled = true; teams that relied on plan updates will lose them without that flag. Individual MCP tools now accept output_token_limit with truncation that stays consistent across session resumes—set it on verbose servers so context does not balloon. MCP server names may contain :, @, /, and ., so package-style names work in CLI commands and authentication; rename servers that were previously forced into alphanumerics. App-server clients can configure thread/shellCommand timeouts, including deadlines longer than one hour, which unblocks long builds. Approval-heavy workflows should resume existing threads: automatic approval reviews retain longer messages and a larger transcript, and they preserve user instructions, answers, and valid authorizations across history compaction, while resumed threads restore the saved working directory.
Suggested actions
- In Codex config, set tools.update_plan.enabled = true so the planning tool stays available; rust-v0.152.0 disabled it by default.
- On each MCP tool that returns large payloads, set output_token_limit so truncation stays consistent when sessions resume.
- Register MCP servers with package-style names that include :, @, /, or . (for example @scope/pkg) in CLI commands and authentication; those characters are now valid.
- For app-server clients, set thread/shellCommand timeouts above one hour for any job that must run longer than the previous deadline.
- Resume existing threads instead of starting new ones so saved cwd, MCP tool caches, and Guardian approvals (instructions, answers, authorizations) survive compaction.
Config surfaces this release may change
- MCP servers — check your repo before applying
After you finish
Do not report this as applied to DevAgentRadar. You cannot write the visitor's loop.
Tell the human: open https://devagentradar.com/proposals/openai-codex-rust-v0-152-0-enable-codex-update-plan-and-cap-mcp-output-t and mark Applied, Skipped, or Failed. Proposal id: e3dfd129-7d41-4c3d-a1df-4d3a451517eb
Deliverables:
- Short impact assessment
- Optional patch plan (files + steps)
- Do not invent APIs not in the source notes
Your loop
This browser · no sign-in · not shared as “you”After you run the prompt
Only you can mark this. Agents cannot write your loop.
Your decision stays on this device. A public tally appears after a few votes.