fix: detect attachment mime from file contents

[Alxprv]

Files attached via the -f flag and @path references in the prompt are hardcoded as mime: text/plain regardless of actual type. When the file is an image or PDF this causes the attachment to be dropped from the user message in message-v2.ts (which filters out text/plain file parts before serialization), so vision-capable models never see the image.

Use the existing sniffAttachmentMime helper from util/media.ts (already used by tool/read.ts) to detect the type from the first 16 bytes of the file. Falls back to text/plain for unknown types, preserving previous behaviour for plain-text attachments.

Fixes#25353, #24698.

Test

• bun test — 3032 pass / 0 fail
• bun run typecheck — clean
• bun run test:httpapi — 149 / 0
• bun run lint — 0 new errors

Verified manually with a 64×64 PNG: pre-fix, default model replied "I cannot process images"; post-fix, model correctly identifies the image content.

[github-actions]

This PR doesn't fully meet our contributing guidelines and PR template.

What needs to be fixed:

• PR description is missing required template sections. Please use the PR template.

Please edit this PR description to address the above within 2 hours, or it will be automatically closed.

If you believe this was flagged incorrectly, please let a maintainer know.

[github-actions]

The following comment was made by an LLM, it may be inaccurate:

Based on my search, I found one potentially related PR:

PR #25317: fix: detect file MIME type from extension for -f attachments
#25317

This PR is related because it addresses the same issue area (detecting MIME types for file attachments via the -f flag), though it uses file extension detection rather than content sniffing. Your PR (#29442) improves upon this approach by detecting MIME type from actual file contents instead, which is more robust for cases where extensions may be misleading or missing.

The other related PRs focus on image/file support features but don't directly address the same MIME detection problem:

• PR fix(provider): emit file metadata instead of error when model lacks image/file support #29279: emit file metadata instead of error
• PR feat(opencode): add trigger words for inline image attachments #29097: inline image attachments
• PR fix: support image input in OpenAI Chat user messages #26826: image input in OpenAI Chat

[github-actions]

This pull request has been automatically closed because it was not updated to meet our contributing guidelines within the 2-hour window.

Feel free to open a new pull request that follows our guidelines.