fix(opencode): report skipped plugins when no server entrypoint is found

[AllanSantos-DV]

Issue for this PR

Closes #44367

Type of change

• Bug fix

What does this PR do?

In the server plugin loader, the missing report callback was empty — an npm plugin whose package has no detectable server entrypoint was dropped with zero feedback, unlike install/compatibility/load failures which call publishPluginError. This fills it in with the same pattern:

Plugin <spec> skipped: Plugin <spec> does not expose a server entrypoint

How did you verify your code works?

• 3-line change inside the existing report object; no behavior change for successfully loaded plugins.
• tsc --noEmit: only pre-existing errors in packages/tui/ (unrelated to this file).
• Repro scenario for the underlying silence is documented in [Desktop][Windows] npm plugin from config is silently never loaded — no install attempt logged, docs/cache mismatch #44367.

Checklist

• I have tested my changes locally
• I have not included unrelated changes in this PR

[github-actions]

Thanks for your contribution!

This PR doesn't have a linked issue. All PRs must reference an existing issue.

Please:

1. Open an issue describing the bug/feature (if one doesn't exist)
2. Add Fixes #<number> or Closes #<number> to this PR description

See CONTRIBUTING.md for details.

[github-actions]

Thanks for updating your PR! It now meets our contributing guidelines. 👍

[Enough1122]

AI code review — automated review for reference, author can ignore or act on any point.

Right fix for a real gap — previously a plugin whose entrypoint couldn't be resolved vanished silently, and this closes that while keeping the wording consistent with the existing compatibility-stage message (packages/opencode/src/plugin/index.ts:380). I checked the loader side: report.missing fires exactly once per candidate and returns { retry: false } (packages/opencode/src/plugin/loader.ts:~301-321), so there's no duplicate-toast risk across retries, and it's correctly bypassed when the missing resolver hook recovers a value (e.g. TUI-only packages).

Two small suggestions:

1. Make the message actionable. At packages/opencode/src/plugin/index.ts:192-194 you drop both the retry flag and the Missing payload the loader passes as the fourth argument. Including the plugin origin helps users act, e.g. `Plugin ${candidate.plan.spec} (${candidate.origin}) skipped: ${message}` — otherwise for config-sourced vs local-file plugins the report reads identically despite very different fixes.

2. No regression test. The change is a pure side-effect on a report callback; a small assertion in packages/opencode/test/plugin/loader-shared.test.ts that a candidate failing resolution triggers publishPluginError with the skipped wording would pin both the wiring and the copy.

Otherwise good to merge.

[AllanSantos-DV]

Thanks for the careful review! Both points applied in e0d30fd:

1. Skip reports now carry the declaring origin: Plugin acme-plugin@1.0.0 (local: /path/to/config-dir) skipped: ...
2. Formatting extracted to plugin/report.ts, with a regression test (loader-shared.test.ts) pinning both the wiring and the exact skipped wording.

[Enough1122]

Both points look good - carrying the declaring origin into skip reports makes them actionable, and extracting the formatting behind regression tests keeps it stable. Verified e0d30fd covers both. Nothing further from me.

[github-actions]

Automated PR Cleanup

Thank you for contributing to opencode.

Due to the high volume of PRs from users and AI agents, we periodically close older PRs using automated criteria so maintainers can focus review time on the most active and community-supported contributions.

This PR was closed because it matched the following cleanup criteria:

• The PR was created more than 1 month ago
• The PR had fewer than 2 positive reactions
• Positive reactions are counted as thumbs-up, heart, celebration, or rocket reactions on the PR

PRs created within the last month are not affected by this cleanup.

If you believe this PR was closed incorrectly, or if you are still actively working on it, please leave a comment explaining why it should be reopened. A maintainer can review and reopen it if appropriate.

Thanks again for taking the time to contribute.