fix(skills): name stale index entries instead of generic fetch failure

[nikkoxgonzales]

What does this PR do?

_resolve_source_meta_and_bundle already distinguishes index-hit-without-files from unknown identifiers, but do_install printed the same generic 'Could not fetch' for both — sending users off to re-check spellings for what is actually a stale skills.sh entry. Split the message (stale branch names the index + suggests search), and add a staleness caveat to do_search results from skills.sh.

Related Issue

Fixes #3259
Supersedes #3261 (stale since July — re-applied onto the current _print_fetch_failure helper, with credit to Mibayy)

Type of Change

• 🐛 Bug fix (non-breaking change that fixes an issue)

Changes Made

• hermes_cli/skills_hub.py: _print_fetch_failure takes meta/source and branches stale vs unknown; do_install passes them; do_search footer warns on skills.sh results
• tests/hermes_cli/test_skills_hub.py: 3 tests (stale message, generic preserved, search caveat)

How to Test

1. pytest tests/hermes_cli/test_skills_hub.py -q → 11 passed
2. Install a skills.sh identifier whose files 404 → 'Stale index entry' naming skills-sh; unknown identifier → original generic message

Checklist

Code

• I've read the Contributing Guide
• My commit messages follow Conventional Commits
• I searched for existing PRs (fix(skills): clear error message for stale skills.sh index entries #3261 stale/superseded, noted above)
• My PR contains only changes related to this fix
• I've added tests for my changes
• I've tested on my platform: Windows 11
• ruff check on both changed files passes

Documentation & Housekeeping

• N/A (no docs/config/schema changes; CLI output text only)

[Enough1122]

AI code review — automated review for reference; please use your judgment.

Summary

Names stale skills.sh index entries explicitly at install time and adds a search-time caveat, instead of reporting a generic fetch failure. Small, focused UX fix with tests for both the stale and unknown-identifier paths.

Findings

Non-blocking (nit) — hermes_cli/skills_hub.py:29: src_id resolution calls getattr(source, "source_id", lambda: "the registry")(). If source is None this safely falls back, but if any source exposes source_id as a plain string attribute rather than a method, the trailing () raises TypeError. Consider guarding with callable(...):

sid = getattr(source, "source_id", None)
src_id = sid() if callable(sid) else (sid or "the registry")

Non-blocking (nit) — hermes_cli/skills_hub.py:11: any(r.source in (...) for r in results) iterates results after the table was already printed from it. This is fine if results is a list, but if it ever becomes a one-shot generator the note would silently never print. No action needed if the caller always passes a list — just flagging the assumption.

Tests

New tests cover stale-entry messaging (test_do_install_stale_index_names_the_problem), the unchanged generic path (test_do_install_unknown_identifier_stays_generic), and the search caveat (test_do_search_warns_about_skills_sh_staleness). Good coverage for the change size.

Verdict

Looks good. The two nits above are optional hardening; nothing blocking.

[teknium1]

Merged via #108906 — your commit is on main as 257a704d18d1 with authorship preserved. One follow-up on top (08bb58bd4a58): the stale-entry verdict is only emitted when no adapter was rate limited (a throttled fetch also yields meta-without-bundle), and the per-search caveat was dropped since the install error now names the condition. Thanks; #3259 is closed.