What does this PR do?

hermes curator restore <skill> was failing with skill '<name>' not found in archive whenever the archived skill lived under a nested category subdirectory (e.g. .archive/openclaw-imports/<skill>/, .archive/hermes-agent/<skill>/). The skill directory existed on disk but was invisible to the lookup because restore_skill() walked only the top level of .archive/.

tools/skill_usage.py:389,392 used archive_root.iterdir() for both the exact-name and prefix-match candidate scans. Switching both to archive_root.rglob("*") makes the lookup descend into nested subdirs while preserving the existing is_dir() and p.name == skill_name (and prefix) filter, so the scan still picks up only matching skill directories.

Related Issue

Fixes #17942

Type of Change

• 🐛 Bug fix (non-breaking change that fixes an issue)
• ✨ New feature (non-breaking change that adds functionality)
• 🔒 Security fix
• 📝 Documentation update
• ✅ Tests (adding or improving test coverage)
• ♻️ Refactor (no behavior change)
• 🎯 New skill (bundled or hub)

Changes Made

• tools/skill_usage.py — restore_skill(): switched the exact-match and prefix-match candidate scans from archive_root.iterdir() to archive_root.rglob("*") so nested archive subdirectories are searched.
• tests/tools/test_skill_usage.py — added test_restore_skill_finds_nested_archive_subdir (exact match under a nested category) and test_restore_skill_finds_nested_timestamped_prefix (timestamped-dupe prefix match under a nested category) to lock in the recursive-walk behaviour.

How to Test

1. Place an archived skill under a nested category, e.g.:

   ~/.hermes/skills/.archive/openclaw-imports/cognitive-empathy/SKILL.md
   

2. Run hermes curator restore cognitive-empathy — before this patch this returned "not found in archive"; after, the skill is moved back to ~/.hermes/skills/cognitive-empathy/ and its state flips to active.
3. pytest tests/tools/test_skill_usage.py -q — 37 tests pass (35 prior + 2 new).

Checklist

Code

• I've read the Contributing Guide
• My commit messages follow Conventional Commits (fix(scope):, feat(scope):, etc.)
• I searched for existing PRs to make sure this isn't a duplicate
• My PR contains only changes related to this fix/feature (no unrelated commits)
• I've run pytest tests/ -q and all tests pass
• I've added tests for my changes (required for bug fixes, strongly encouraged for features)
• I've tested on my platform: macOS 15 (Darwin 23.2)

Documentation & Housekeeping

• I've updated relevant documentation (README, docs/, docstrings) — N/A (no user-visible API change; the docstring on restore_skill already covers behaviour)
• I've updated cli-config.yaml.example if I added/changed config keys — N/A
• I've updated CONTRIBUTING.md or AGENTS.md if I changed architecture or workflows — N/A
• I've considered cross-platform impact (Windows, macOS) per the compatibility guide — rglob is pathlib-portable; tests use tmp_path and avoid OS-specific paths
• I've updated tool descriptions/schemas if I changed tool behavior — N/A