[echarrod]

Description

Refs #2654. Follow-up to #4919 (Claude) and #5116 (Droid).

Adds a Copilot-only integration option:

specify init my-project --integration copilot --integration-options="--no-model-invocation"

• In the default skills layout it writes disable-model-invocation: true on the core speckit-* command skills, including preset overrides of them. VS Code's agent skills docs say this requires manual invocation through the / slash command only. The default is unchanged (the key is omitted, which VS Code treats as false).
• Extension skills stay model-invocable, because mandatory hooks such as git's speckit.git.feature need the model to invoke them.
• The option persists through integration.json, so specify integration upgrade copilot reapplies it.

Differences from Droid:

• Copilot has two layouts. Under --commands the files are .agent.md and .prompt.md, not skills, so the option is ignored and no flag is written.
• Only the VS Code docs list the field; GitHub's Copilot CLI skills docs do not. In a manual check with Copilot CLI 1.0.70 the model called its skill tool for speckit-specify without the flag and did not with it, and an explicit /speckit-specify still ran. With the flag the model could still read the skill's SKILL.md as a plain file, so it is not a hard block. Neither source says whether a flagged skill keeps its description out of the model's context, so the context-budget benefit from Allow opting out of model invocation for installed skills (large monorepo skill-budget pain) #2654 is unverified.
• Copilot is not multi_install_safe; the existing multi-install note is unchanged.
• The frontmatter injection preserves CRLF line endings, as Claude's does.
• Copilot already has other options, so the docs add a section rather than editing a "no options" line.

Only the Copilot package, its tests and docs/integrations/copilot.md change; no shared base.py change.

AI Disclosure

• I did not use AI assistance for this contribution
• I did use AI assistance (fill in the disclosure below)

AI disclosure: Written for @echarrod by Claude Code (claude-sonnet-5-5) running autonomously. It wrote the code, tests and docs, ran the test suite and an automated /code-review medium. @echarrod reviewed it before marking it ready.