feat: leave out forks and excluded repos

Forks are skipped automatically unless overrides.yaml sets include: true
(mistral-vibe keeps its documentation site listed); exclude: true hides
a repo for good. Removes arc42-generator (fork) and brownfield-experiment
from the overview and llms.txt.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

Author: rdmueller

VERSION

@@ -1 +1 @@
-0.2.0
+0.3.0

generate.py

@@ -59,9 +59,14 @@ def is_http(url):
     return urlparse(url).scheme in ("http", "https")
 
 
-def select_projects(repos, org_site_repo=ORG_SITE_REPO):
-    chosen = [r for r in repos if r.get("has_pages") and not r.get("archived")
-              and r["name"].lower() != org_site_repo.lower()]
+def select_projects(repos, overrides, org_site_repo=ORG_SITE_REPO):
+    """Repos with Pages, minus archived, the org site, excluded ones and forks (unless included)."""
+    def wanted(r):
+        ov = overrides.get(r["name"], {})
+        return (r.get("has_pages") and not r.get("archived")
+                and r["name"].lower() != org_site_repo.lower()
+                and not ov.get("exclude") and (not r.get("fork") or ov.get("include")))
+    chosen = [r for r in repos if wanted(r)]
     return sorted(chosen, key=lambda r: r["name"].lower())
 
 
@@ -76,8 +81,13 @@ def description(repo, overrides):
 
 def load_overrides(path):
     data = yaml.safe_load(path.read_text(encoding="utf-8")) if path.exists() else None
-    return {str(k): {f: str(v[f]) for f in ("name", "description") if v and v.get(f)}
-            for k, v in (data or {}).items()}
+    result = {}
+    for key, v in (data or {}).items():
+        v = v or {}
+        entry = {f: str(v[f]) for f in ("name", "description") if v.get(f)}
+        entry.update({f: True for f in ("include", "exclude") if v.get(f) is True})
+        result[str(key)] = entry
+    return result
 
 
 def load_yaml(path):
@@ -305,8 +315,8 @@ def main():
     args = ap.parse_args()
 
     all_repos = gh_api_list(f"orgs/{args.org}/repos?per_page=100")
-    repos = select_projects(all_repos)
     overrides = load_overrides(ROOT / "overrides.yaml")
+    repos = select_projects(all_repos, overrides)
     links = load_yaml(ROOT / "links.yaml")
     # Archived repos keep serving their Pages, so check against every repo name
     validate_links(links, [r["name"] for r in all_repos])

overrides.yaml

@@ -1,17 +1,23 @@
 # Per-repository overrides for the overview page and llms.txt.
-# Keys are repository names. Both fields are optional and win over the GitHub values,
-# so the daily regeneration keeps them. URL and repository link stay unchanged.
+# Keys are repository names. All fields are optional; name and description win over
+# the GitHub values, so the daily regeneration keeps them. URL and repository link stay unchanged.
+# Forks are left out automatically unless they set include: true.
 #
 # <repo>:
 #   name: Display name
 #   description: Short description
+#   include: true   # list this repo although it is a fork
+#   exclude: true   # never list this repo
 
 mistral-vibe:
   name: Mistral Vibe Doku
+  include: true  # a fork, but the documentation site is ours
   description: Documentation of Mistral Vibe, the minimal CLI coding agent – specification, arc42 architecture and ADRs, reverse-engineered from the source code.
 Semantic-Anchors:
   description: Shared vocabulary for LLM communication – curated semantic anchors and semantic contracts for precise prompts.
 vibe-coding-risk-radar:
   description: Risk framework for AI-generated code – five dimensions mapped to four risk tiers, with Claude Code skills for assessment and mitigation.
 ultrathink.club:
   description: An AI think-tank community. Coming soon.
+brownfield-experiment:
+  exclude: true

tests/test_generate.py

@@ -10,19 +10,33 @@
 ORG = "LLM-Coding"
 
 
-def repo(name, desc="", pages=True, archived=False, homepage=""):
+def repo(name, desc="", pages=True, archived=False, homepage="", fork=False):
     return {"name": name, "description": desc, "has_pages": pages,
-            "archived": archived, "homepage": homepage,
+            "archived": archived, "homepage": homepage, "fork": fork,
             "html_url": f"https://github.com/{ORG}/{name}"}
 
 
 def test_select_projects_filters_pages_archived_and_org_site():
     repos = [repo("b"), repo("a"), repo("no-pages", pages=False),
              repo("old", archived=True), repo("llm-coding.github.io")]
-    names = [r["name"] for r in g.select_projects(repos, "llm-coding.github.io")]
+    names = [r["name"] for r in g.select_projects(repos, {}, "llm-coding.github.io")]
     assert names == ["a", "b"]
 
 
+def test_select_projects_drops_forks_and_excluded_unless_included():
+    repos = [repo("own"), repo("fork", fork=True), repo("kept-fork", fork=True),
+             repo("hidden")]
+    ov = {"kept-fork": {"include": True}, "hidden": {"exclude": True}}
+    assert [r["name"] for r in g.select_projects(repos, ov)] == ["kept-fork", "own"]
+
+
+def test_load_overrides_reads_flags(tmp_path):
+    f = tmp_path / "o.yaml"
+    f.write_text("a:\n  name: A\n  exclude: true\nb:\n  include: true\n")
+    assert g.load_overrides(f) == {"a": {"name": "A", "exclude": True},
+                                   "b": {"include": True}}
+
+
 def test_overrides_win_over_github_for_name_and_description():
     ov = {"x": {"name": "X Docs", "description": "Override"}}
     assert g.display_name(repo("x"), ov) == "X Docs"