fix: stop misclassifying plugin skills as sandbox-only via stale cache

[Rain-0x01-39]

Fixes #9754

Modifications / 改动点

• SkillManager.is_sandbox_only_skill() now returns False when the name resolves to a plugin-provided skill directory, before consulting the sandbox skills cache.

• Previously, the method only checked the local skills_root and data/sandbox_skills_cache.json. That cache is persistent and never cleaned up when switching back to the local runtime, so a leftover entry with the same name as a builtin-star skill (e.g. the pdf skill shipped by the builtin astrbot star) made SkillsService.resolve_local_skill_dir() reject the skill with PermissionError: Sandbox preset skill cannot be opened from local skill files. even though the skill has real local files under astrbot/builtin_stars/astrbot/skills/ and should be viewable read-only in the Dashboard.

• Added a regression test test_plugin_skill_with_stale_sandbox_cache_is_not_sandbox_only.

• This is NOT a breaking change. / 这不是一个破坏性变更。

Screenshots or Test Results / 运行截图或测试结果

$ uv run pytest tests/test_skill_manager_sandbox_cache.py -q
4 passed, 1 warning in 9.41s

The new test fails on master (assertion at tests/test_skill_manager_sandbox_cache.py:211) and passes with the fix. Other skill-related suites show no new failures (9 pre-existing Windows path-related failures in tests/test_skill_metadata_enrichment.py etc. are identical with and without this change).

uv run ruff format and uv run ruff check pass on the touched files.

---

Checklist / 检查清单

• 😊 If there are new features added in the PR, I have discussed it with the authors through issues/emails, etc.
  / 如果 PR 中有新加入的功能，已经通过 Issue / 邮件等方式和作者讨论过。

• 👀 My changes have been well-tested, and "Verification Steps" and "Screenshots" have been provided above.
  / 我的更改经过了良好的测试，并已在上方提供了“验证步骤”和“运行截图”。

• 🤓 I have ensured that no new dependencies are introduced, OR if new dependencies are introduced, they have been added to the appropriate locations in requirements.txt and pyproject.toml.
  / 我确保没有引入新依赖库，或者引入了新依赖库的同时将其添加到 requirements.txt 和 pyproject.toml 文件相应位置。

• 😮 My changes do not introduce malicious code.
  / 我的更改没有引入恶意代码。

Summary by Sourcery

Correct sandbox classification so plugin-provided skills take precedence over stale sandbox cache entries.

Bug Fixes:

• Prevent plugin-provided skills from being misclassified as sandbox-only when a stale sandbox cache entry has the same name.
• Allow local resolution and read-only dashboard access to plugin skills despite obsolete sandbox cache data.

Tests:

• Add a regression test covering plugin skills that conflict with stale sandbox cache entries.

[sourcery-ai]

Hey - I've reviewed your changes and they look great!

---

Sourcery is free for open source - if you like our reviews please consider sharing them ✨

• X
• Mastodon
• LinkedIn
• Facebook

_{Help me be more useful! Please click 👍 or 👎 on each comment and I'll use the feedback to improve your reviews.}

[Rain-0x01-39]

The only failing check (Run pytest suite) is unrelated to this PR:

FAILED tests/test_anthropic_kimi_code_provider.py::test_create_http_client_uses_anthropic_httpx_module
AttributeError: module 'anthropic._base_client' has no attribute 'httpx'. Did you mean: 'httpx2'?

This test monkeypatches the private attribute anthropic._base_client.httpx, which was renamed to httpx2 in a newer anthropic SDK release that CI resolved during this run (uv.lock pins 0.105.2, where it still passes — verified locally: 27 passed). All other checks including the new regression test pass (2163 passed).