Document attachment support in gh skill

[BagToad]

Related implementation: #14255

Description

The installable gh agent skill does not describe the new repeatable --attach flag. Agents using the skill would miss the supported commands, argument format, Markdown rewriting, video behavior, incompatible modes, authentication requirements, and partial-failure contract.

This adds one concise attachment section to skills/gh/SKILL.md, following the skill's existing bullet-oriented style and inline command examples.

How did you test this change?

I previewed the installable skill package with gh skill publish --dry-run . and read the rendered source alongside the command help and attachment tests on the parent branch.

Key points

• Covers all six issue, pull request, and comment commands that support attachments.
• Keeps invocation guidance in the gh skill rather than documenting implementation details.
• Distinguishes image alt text, standalone video players, inline video links, and unsupported reference-style video images.
• Records the host, token, permission, conflict, and partial-failure constraints agents need before invoking the flag.

Notes for reviewers

Review the new Attaching images and videos section in skills/gh/SKILL.md. The file path is the tracked skill source installed by gh skill install cli/cli gh --scope user.

Authorship and follow-up

Who wrote this:

• A human wrote it.
• An agent wrote it under close human direction.
• An agent wrote it independently, and no human has guided the implementation beyond the initial prompt.

Who answers review comments:

• @BagToad will read and reply directly. Name the account.
• An agent will draft replies and @username will read them before they are posted.
• Nobody has explicitly committed to replying.

[babakks]

Thanks for capping off the stack with the docs, @BagToad! 📚

The attachment section reads clearly and covers the important edges: supported file types, alt text syntax and shell quoting, the reference-rewrite versus append behavior, the video limitations, the per-command conflicts, and the partial-upload exit semantics. Nice and thorough.

I left one small suggestion: it would help to spell out that the filesystem paths in the body markdown and in --attach resolve either as absolute paths or relative to the directory where gh runs. Non-blocking.

Nice work landing the whole stack! 🎉