Copilot CLI 1.0.89 Adds .claude/rules Support and GPT-6 Picker

Version 1.0.89 of GitHub Copilot CLI showed up in the project's changelog on September 28. The entry, opened by a teammate rather than pulled fresh from the raw file, lists four things: support for .claude/rules, picker entries for GPT-6 Sol and Luna, support for claude-opus-5.5, and a set of MCP and UI fixes. Short list. Real consequences for anyone scripting around Copilot CLI's model selection or config loading.

The .claude/rules item matters because it signals Copilot CLI reading configuration files originally built for Anthropic's Claude Code tooling. If that holds, a team maintaining separate rule files for separate CLI tools could consolidate them, which changes how engineering managers budget time for tooling maintenance. Policy watchers should note the cross-compatibility angle: one vendor's CLI adopting another vendor's config format is the kind of quiet interoperability move that shapes vendor lock-in arguments long before any antitrust filing mentions it.

The GPT-6 Sol and Luna picker items are listed without elaboration. The changelog does not explain whether Sol and Luna are distinct model sizes, context-window variants, or release channels, and the entry gives no pricing or availability detail. Claude-opus-5.5 support is stated plainly as added, but again without a version note on what it replaces or supplements. Anyone budgeting API spend or testing output quality needs to treat these as model names confirmed present in the picker, not as confirmed behavior, cost, or default status.

Workers who use Copilot CLI daily get a concrete decision to make: whether to pin a model version now or wait, given that a picker entry is not the same as a stable default. Managers overseeing a team's AI tooling spend should ask whether claude-opus-5.5 usage bills differently than prior Claude models routed through Copilot, since the changelog says nothing about cost. Policy watchers tracking vendor interoperability have a small but citable data point: a major CLI tool reading a competitor's config directory format, which is the kind of technical detail that outlasts any press statement about partnership or competition.

The mechanism worth checking is provenance. A teammate opened this changelog entry rather than the raw file being re-extracted at publication time, so the four listed items reflect what one person read and recorded on September 28, not a verified diff against the actual release artifact. That gap is small until it is not. MCP and UI fixes are mentioned with zero specifics, which means anyone relying on Model Context Protocol integrations should check the actual release notes or binary version string before assuming a particular bug is resolved.

The next move is mechanical: pull the live changelog.md from the copilot-cli repository and diff it against what circulated on September 28, because a teammate's read of a raw file is not the same as the file itself. Until that check happens, treat claude-opus-5.5 and the GPT-6 Sol and Luna picker entries as reported, not confirmed, and hold off on any workflow that assumes their behavior is final.