Gemini CLI 0.64 preview moves settings to V2: recheck Sume
The 0.64.0 preview moves settings from V1 to V2. After you upgrade, confirm the Sume hosted MCP server still connects with a free read before any paid call.

After installing the Gemini CLI 0.64.0 preview, check that your Sume hosted MCP server still connects before you run anything paid. The GitHub release notes dated October 6, 2026 list a settings migration from V1 to V2 among the major updates. The notes do not say what it does to an mcpServers entry, so assume nothing and verify.
The check is cheap. Call a free Sume read such as mcp_health or tools_list and look at the result.
What the recent releases list
From the GitHub releases page; items not related to MCP or settings are left out.
| Version | Date | Relevant item |
|---|---|---|
| 0.65.0 nightly | Oct 8, 2026 | Fixes aimed at infinite verification and OAuth retry loops |
| 0.64.0 preview | Oct 6, 2026 | Settings migration from V1 to V2; atomic file tool writes; resumed session history no longer deleted |
| 0.63.0 stable | Oct 6, 2026 | Auth loop fix for file contention and headless keyring; MCP config validation separates missing enablement from malformed JSON |
Upgrade steps
Before upgrading, copy your settings file somewhere safe. After upgrading, open the settings and find the Sume entry; if it moved or lost a field, restore it from the copy. Then connect to https://mcp.sume.com/mcp and run mcp_health. With an API key, the result should show the key as the auth source; with OAuth, expect the browser sign-in the first time.
- Keep the Sume key in an environment variable, not in the settings file.
- If the server shows as not enabled, check the enablement file as well as the settings file; 0.63 reports those as separate problems.
- Run a
dry_run=truepreview before the first paid call after any upgrade.
A migration checklist for a team
If several people share a configuration, make one person the canary. They upgrade first, run the free read, and report what changed in the file. Write down the diff of the settings file before and after, because a V1 to V2 move can rename keys and you want to see which. Only then roll the preview to everyone. The preview label means it is not the stable line, so keep a pinned stable install on at least one machine for automation that runs unattended.
A fallback if the entry broke
If the Sume server no longer appears, re-add it from your copy rather than hand-editing around the migration. If the key was stored in the file, move it to an environment variable at the same time. If the CLI still says the server is not enabled, remember the 0.63 release separated a missing enablement file from malformed JSON, so look at both before you decide the entry is wrong.
What Sume does not do
Sume does not know which CLI version you run. It sees only the HTTP request. If a settings migration drops the Authorization header, the failure shows up as an authentication error from the Sume side, not as a version error. Under OAuth, a Sume access token lasts 3600 seconds and has no refresh token, so a sign-in prompt an hour later is expected rather than a regression.
Sources
Related posts
More in Integrations
- Gemini CLI httpUrl or url for Sume MCP: which key to use
In Gemini CLI settings, url is the SSE endpoint and httpUrl is HTTP streaming. Sume's endpoint is streamable HTTP, so use httpUrl. Config and checks inside.
- Gemini CLI includeTools for Sume MCP: expose reads, keep trust false
Gemini CLI can allowlist MCP tools with includeTools, and trust defaults to false. Use both to give an agent Sume reads without paid generation.
- Gemini CLI MCP timeout is 600000 ms; Sume jobs_wait caps at 55 s
Gemini CLI waits up to 10 minutes per MCP request, but Sume cuts jobs_wait at 55 seconds. Set the client timeout low and loop on wait_slice_expired.
- Can GLM-5.3 call Sume's hosted MCP? Function calling, not MCP
Z.ai's GLM-5.3 page lists function calling and does not mention MCP. How a GLM agent reaches Sume through an MCP client or through plain HTTP.
Written by Sume