fix(github-research): disclose truncated trees (#7815)

Co-authored-by: Willem Jiang <willem.jiang@gmail.com>

README.md

@@ -1913,6 +1913,7 @@ File reads accept raw repository-relative paths and preserve spaces, Unicode,
 `#`, `?`, and literal percent signs with either HTTP client.
 Directory listings ask for the repository's own default branch, so a repository that
 uses neither `main` nor `master` reports its current tree instead of a stale branch.
+They indicate when GitHub or the local output limit truncates the listing.
 The `issues` command excludes pull requests and follows additional pages to fill
 its requested issue limit; the `prs` command remains available for pull requests.
 

skills/public/github-deep-research/SKILL.md

@@ -57,6 +57,7 @@ python /path/to/skill/scripts/github_api.py <owner> <repo> tree
 - releases
 
 The `issues` command returns repository issues only, excluding pull requests. It follows additional pages when needed to fill the requested issue limit; use `prs` to research pull requests separately.
+The `tree` command marks results truncated by GitHub or by its own output limit; inspect subdirectories separately before drawing conclusions from an incomplete tree.
 
 **Round 2 - Discovery (3-5 web_search)**
 - Get overview and identify key terms

skills/public/github-deep-research/scripts/github_api.py

@@ -252,7 +252,12 @@ class GitHubAPI:
                 else:
                     lines.append(f"{indent}{name}")
 
-        return "\n".join(lines[:100])  # Limit output
+        output = lines[:100]
+        if len(lines) > 100:
+            output.append("[Tree output limit: additional entries omitted]")
+        if tree_data.get("truncated"):
+            output.append("[GitHub truncated the tree; additional entries may be missing]")
+        return "\n".join(output)
 
     def summarize_repo(self, owner: str, repo: str) -> Dict:
         """

tests/skills/test_github_deep_research.py

@@ -173,6 +173,35 @@ def test_tree_does_not_guess_another_branch(monkeypatch, fallback_client):
     assert asked == ["/repos/owner/repo/git/trees/HEAD"]
 
 
+@pytest.mark.parametrize("truncated", [False, True])
+def test_tree_reports_upstream_truncation(fallback_client, truncated):
+    client, _ = fallback_client
+    result = client.format_tree({"tree": [{"path": "README.md", "type": "blob"}], "truncated": truncated})
+    assert result.startswith("README.md")
+    assert ("GitHub" in result and "truncated" in result) is truncated
+
+
+@pytest.mark.parametrize("count", [99, 100, 101])
+def test_tree_reports_local_entry_limit(fallback_client, count):
+    client, _ = fallback_client
+    tree = [{"path": f"file-{i}", "type": "blob"} for i in range(count)]
+    result = client.format_tree({"tree": tree, "truncated": False})
+    assert ("output limit" in result) is (count > 100)
+    assert "file-98" in result
+    if count > 100:
+        assert "file-100" not in result
+
+
+def test_tree_reports_both_limits_and_ignores_entries_below_max_depth(fallback_client):
+    client, _ = fallback_client
+    hidden = [{"path": f"folder/file-{i}", "type": "blob"} for i in range(101)]
+    assert client.format_tree({"tree": hidden}, max_depth=1) == ""
+    visible = [{"path": f"file-{i}", "type": "blob"} for i in range(101)]
+    result = client.format_tree({"tree": visible, "truncated": True})
+    assert "output limit" in result
+    assert "GitHub truncated" in result
+
+
 def _client_returning_status(monkeypatch, transport, status, body=b'{"ok": true}'):
     """One client whose transport answers ``status`` for every request, plus its module.