feat(skills): install skills from a direct HTTP(S) URL (#16323)

* feat(skills): install skills from a direct HTTP(S) URL

Adds UrlSource adapter so `hermes skills install <url-to-SKILL.md>` and
`/skills install <url>` work as first-class operations — no more
improvising with curl + patch + cp.

- Claims identifiers that start with http(s):// and end in .md
- Skips /.well-known/skills/ URLs (WellKnownSkillSource handles those)
- Skill name from YAML frontmatter, URL-slug fallback
- Single-file SKILL.md only (v1 scope — multi-file skills need a manifest)
- Trust level 'community'; full security scan still runs
- Lock file stores the URL as identifier so `hermes skills update`
  re-fetches from the same URL cleanly

Scope matches real user need from @versun's docx feedback where
`https://sharethis.chat/SKILL.md` had no first-class install path.

* feat(skills): interactive name/category for URL installs + --name override

Follow-up to the UrlSource adapter. The previous commit fell back to weak
heuristics when frontmatter had no ``name:`` and could produce garbage names
like ``SKILL`` or ``unnamed-skill``. Now:

tools/skills_hub.py
- ``UrlSource._is_valid_skill_name()`` — strict identifier check
  (``^[a-z][a-z0-9_-]*$``), rejects sentinel values (``SKILL``, ``README``,
  ``INDEX``, ``unnamed-skill``, empty, non-strings).
- ``_resolve_skill_name()`` returns ``Optional[str]`` — ``None`` when
  nothing valid is resolvable. Also ignores unsafe frontmatter names
  (``../evil``) and falls through to URL slug instead of returning None
  immediately, so a URL with a bad frontmatter but a good path still
  works.
- ``fetch()``/``inspect()`` carry an ``awaiting_name=True`` marker in
  metadata/extra when resolution fails, letting ``do_install`` decide
  whether to prompt, apply an override, or error out.

hermes_cli/skills_hub.py
- ``do_install`` gains a ``name_override`` parameter.
- On URL-sourced bundles with ``awaiting_name=True``:
  1. If ``name_override`` is valid → use it.
  2. If ``name_override`` is invalid → refuse with a clear error.
  3. Else if ``skip_confirm=True`` (non-interactive: slash / TUI /
     gateway / scripts) → refuse with an actionable retry hint pointing
     at ``--name <your-name>`` on both CLI and slash forms.
  4. Else (interactive TTY) → prompt for the name.
- Interactive TTY also prompts for a category when none is given for a
  URL-sourced install, hinting existing category buckets so users can
  reuse ``productivity``, ``devops``, etc. Empty input → flat install.
- ``_existing_categories()`` scans ``~/.hermes/skills/`` for subdirs that
  look like category buckets (contain nested SKILL.md files); skips
  top-level skills and hidden dirs.
- ``_prompt_for_skill_name()`` / ``_prompt_for_category()`` helpers
  (EOF/Ctrl-C-safe, match the existing ``Confirm [y/N]`` prompt style).

hermes_cli/main.py
- ``hermes skills install`` argparse gains ``--name <name>``.

hermes_cli/skills_hub.py (slash)
- ``/skills install <url> --name <x>`` parsing added.

Tests
- tests/tools/test_skills_hub.py: updated ``UrlSource`` tests to assert
  the new ``awaiting_name`` metadata; added 4 new tests for
  ``_is_valid_skill_name`` rejection sets and the awaiting-name marker.
- tests/hermes_cli/test_skills_hub.py: 8 new tests covering --name
  override accept/reject, non-interactive error, interactive name prompt,
  interactive category prompt, cancel-aborts-install, and
  ``_existing_categories`` scan behavior (buckets vs flat skills).
- E2E verified all four paths (no-name/no-override → error;
  --name override → install; frontmatter name → install;
  invalid --name → rejection).

---------

Co-authored-by: teknium1 <teknium@noreply.github.com>
This commit is contained in:
Teknium
2026-04-26 20:57:10 -07:00
committed by GitHub
parent b288934dff
commit 9c416e20ab
5 changed files with 773 additions and 7 deletions

View File

@@ -931,6 +931,176 @@ class WellKnownSkillSource(SkillSource):
return f"well-known:{base_url.rstrip('/')}/{skill_name}"
# ---------------------------------------------------------------------------
# Direct URL source adapter
# ---------------------------------------------------------------------------
class UrlSource(SkillSource):
"""Fetch a single-file SKILL.md skill directly from an HTTP(S) URL.
The identifier IS the URL (e.g. ``https://example.com/path/SKILL.md``).
Only single-file skills are supported — multi-file skills with
``references/`` or ``scripts/`` subfolders need a manifest we can't
discover from a bare URL.
The skill name is read from the ``name:`` field in the SKILL.md YAML
frontmatter (with a URL-slug fallback). Trust level is always
``community`` and the same security scan runs as for every other source.
"""
def source_id(self) -> str:
return "url"
def trust_level_for(self, identifier: str) -> str:
return "community"
# Search is meaningless for a direct URL — skip (return empty).
def search(self, query: str, limit: int = 10) -> List[SkillMeta]:
return []
def _matches(self, identifier: str) -> bool:
"""Return True iff this source should handle ``identifier``.
We claim bare HTTP(S) URLs that end in ``.md`` (typically
``.../SKILL.md``). Wrapped identifiers (``github:``,
``well-known:``, etc.) and ``/.well-known/skills/`` URLs are
left for their respective adapters.
"""
if not isinstance(identifier, str):
return False
ident = identifier.strip()
if not ident.lower().startswith(("http://", "https://")):
return False
# Don't steal well-known URLs.
if "/.well-known/skills/" in ident or ident.rstrip("/").endswith("/index.json"):
return False
# Only claim URLs that look like a markdown file.
try:
path = urlparse(ident).path
except ValueError:
return False
return path.lower().endswith(".md")
def inspect(self, identifier: str) -> Optional[SkillMeta]:
if not self._matches(identifier):
return None
url = identifier.strip()
text = self._fetch_text(url)
if text is None:
return None
fm = GitHubSource._parse_frontmatter_quick(text)
name = self._resolve_skill_name(fm, url)
description = str(fm.get("description") or "")
tags: List[str] = []
metadata = fm.get("metadata", {})
if isinstance(metadata, dict):
hermes_meta = metadata.get("hermes", {})
if isinstance(hermes_meta, dict):
raw_tags = hermes_meta.get("tags", [])
if isinstance(raw_tags, list):
tags = [str(t) for t in raw_tags]
return SkillMeta(
name=name or "",
description=description,
source="url",
identifier=url,
trust_level="community",
path=name or "",
tags=tags,
extra={"url": url, "awaiting_name": name is None},
)
def fetch(self, identifier: str) -> Optional[SkillBundle]:
if not self._matches(identifier):
return None
url = identifier.strip()
text = self._fetch_text(url)
if text is None:
return None
fm = GitHubSource._parse_frontmatter_quick(text)
name = self._resolve_skill_name(fm, url)
# When auto-resolution fails, return a bundle with an empty name and
# ``awaiting_name=True`` in metadata. The install flow (``do_install``)
# either prompts the user on a TTY or refuses with an actionable error
# on non-interactive surfaces. Keep the expensive HTTP fetch's result
# so the caller doesn't have to re-download after picking a name.
skill_name = ""
if name is not None:
try:
skill_name = _validate_skill_name(name)
except ValueError:
logger.warning("URL skill %s produced unsafe skill name: %r", url, name)
return None
return SkillBundle(
name=skill_name,
files={"SKILL.md": text},
source="url",
identifier=url,
trust_level="community",
metadata={"url": url, "awaiting_name": not skill_name},
)
@staticmethod
def _fetch_text(url: str) -> Optional[str]:
try:
resp = httpx.get(url, timeout=20, follow_redirects=True)
if resp.status_code == 200:
return resp.text
except httpx.HTTPError as exc:
logger.debug("UrlSource fetch failed for %s: %s", url, exc)
return None
return None
# Skill names must look like identifiers: lowercase letters/digits with
# optional hyphens/underscores. Blocks dangerous (``../evil``) AND useless
# (``SKILL``, ``README``, empty) candidates before they hit the disk.
_VALID_NAME_RE = re.compile(r"^[a-z][a-z0-9_-]*$")
@classmethod
def _is_valid_skill_name(cls, name: Optional[str]) -> bool:
if not isinstance(name, str):
return False
candidate = name.strip().lower()
if not candidate or candidate in {"skill", "readme", "index", "unnamed-skill"}:
return False
return bool(cls._VALID_NAME_RE.match(candidate))
@classmethod
def _resolve_skill_name(cls, fm: dict, url: str) -> Optional[str]:
"""Pick a skill name from frontmatter or URL.
Returns ``None`` when neither source produces a valid identifier;
callers (CLI ``do_install``) then prompt the user or refuse. Preferring
a clean failure over a useless auto-name like ``SKILL`` or ``unnamed-skill``.
"""
# 1. Frontmatter ``name:`` is authoritative when present and valid.
fm_name = fm.get("name") if isinstance(fm, dict) else None
if isinstance(fm_name, str) and cls._is_valid_skill_name(fm_name):
return fm_name.strip()
# 2. URL-slug heuristic: ``.../<name>/SKILL.md`` → ``<name>``;
# ``.../<name>.md`` → ``<name>``. Validate each candidate.
try:
path = urlparse(url).path
except ValueError:
return None
parts = [p for p in path.split("/") if p]
if parts and parts[-1].lower() == "skill.md" and len(parts) >= 2:
candidate = parts[-2]
if cls._is_valid_skill_name(candidate):
return candidate
if parts:
candidate = re.sub(r"\.md$", "", parts[-1], flags=re.IGNORECASE)
if cls._is_valid_skill_name(candidate):
return candidate
# Nothing usable — let the caller handle it.
return None
# ---------------------------------------------------------------------------
# skills.sh source adapter
# ---------------------------------------------------------------------------
@@ -2931,6 +3101,7 @@ def create_source_router(auth: Optional[GitHubAuth] = None) -> List[SkillSource]
HermesIndexSource(auth=auth), # Centralized index (search + resolved install paths)
SkillsShSource(auth=auth),
WellKnownSkillSource(),
UrlSource(), # Direct HTTP(S) URL to a SKILL.md file
GitHubSource(auth=auth, extra_taps=extra_taps),
ClawHubSource(),
ClaudeMarketplaceSource(auth=auth),