[argus] telegram: scope orphan-artifact auto-present to viewable end-products

_orphan_artifacts was too greedy: it linked ANY new file in outputs/,
including throwaway scripts. Asking for the weather (Qwen writes a
fetch .py into outputs/) would yield a spurious /f/fetch.py link.

Now gated by _ORPHAN_PRESENT_EXTS (.html/.svg/.pdf/.png/.jpg/.gif/.webp) —
viewable deliverables only; code/data/scratch (.py/.json/.csv/.txt/.log)
is ignored. Explicit present_files still presents anything. 3 new tests.

Author: Nichol4s

backend/app/channels/manager.py

@@ -388,13 +388,27 @@ def _resolve_attachments(thread_id: str, artifacts: list[str]) -> list[ResolvedA
     return attachments
 
 
+# [argus patch #10] Only these extensions are auto-presented when the agent
+# writes a file to outputs/ but forgets to present_files it. The rule:
+# auto-present VIEWABLE END-PRODUCTS only (a report, a diagram, an image), and
+# NEVER the means used to produce them (a .py fetch script, a .json blob, a
+# scratch .csv/.txt/.log). E.g. asking for the weather, where Qwen writes a
+# throwaway fetch script into outputs/, must NOT yield a "/f/fetch.py" link.
+# The agent can still explicitly present ANYTHING via present_files — this
+# allowlist only gates the *automatic* rescue path.
+_ORPHAN_PRESENT_EXTS = {".html", ".htm", ".svg", ".pdf", ".png", ".jpg", ".jpeg", ".gif", ".webp"}
+
+
 def _orphan_artifacts(thread_id: str, since: float, already: list[str]) -> list[str]:
-    """[argus patch #10] Files the agent wrote to the outputs dir during this
-    run but did NOT present via present_files. Models sometimes write a file
-    and then paste its contents into chat instead of calling present_files
-    (observed with SVG/HTML). We detect those by mtime so they still get a
-    /f/ link rather than a wall of source. Excludes already-presented paths
-    and the render-and-verify skill's *.screenshot.png sidecars (noise)."""
+    """[argus patch #10] Viewable deliverables the agent wrote to outputs/
+    during this run but did NOT present via present_files. Models sometimes
+    write a file and paste its contents into chat instead (observed with SVG);
+    we rescue those so they get a /f/ link rather than a wall of source.
+
+    Deliberately conservative — only the viewable-end-product extensions in
+    _ORPHAN_PRESENT_EXTS qualify. Code/data/scratch files (.py, .json, .csv,
+    .txt, .log, …) are ignored: they're the means, not the answer. Also skips
+    already-presented files and render-and-verify *.screenshot.png sidecars."""
     try:
         from deerflow.config.paths import get_paths
 
@@ -413,6 +427,8 @@ def _orphan_artifacts(thread_id: str, since: float, already: list[str]) -> list[
             continue
         if f.name in already_names or f.name.endswith(".screenshot.png"):
             continue
+        if f.suffix.lower() not in _ORPHAN_PRESENT_EXTS:
+            continue  # not a viewable end-product — skip (e.g. a fetch .py)
         found.append(_OUTPUTS_VIRTUAL_PREFIX + f.name)
     return found
 

backend/tests/test_channel_file_attachments.py

@@ -562,3 +562,42 @@ def test_format_artifact_text_for_unresolved(self):
         result = _format_artifact_text(["/mnt/user-data/outputs/a.txt", "/mnt/user-data/outputs/b.txt"])
         assert "a.txt" in result
         assert "b.txt" in result
+
+
+class TestOrphanArtifacts:
+    """[argus patch #10] _orphan_artifacts must auto-present only viewable
+    end-products (HTML/SVG/PDF/images), never the means (a .py fetch script,
+    scratch .json/.csv/.txt). This is the weather-query guard: asking for the
+    weather, where the model writes a throwaway fetch.py into outputs/, must
+    not produce a /f/fetch.py link."""
+
+    def _orphans(self, tmp_path, files, since=0.0, already=None):
+        from unittest.mock import MagicMock, patch
+        from app.channels.manager import _orphan_artifacts
+
+        thread_id = "t-orphan"
+        outputs_dir = tmp_path / "threads" / thread_id / "user-data" / "outputs"
+        outputs_dir.mkdir(parents=True)
+        for name in files:
+            (outputs_dir / name).write_text("x")
+        mock_paths = MagicMock()
+        mock_paths.sandbox_outputs_dir.return_value = outputs_dir
+        with patch("deerflow.config.paths.get_paths", return_value=mock_paths):
+            return _orphan_artifacts(thread_id, since, already or [])
+
+    def test_ignores_code_and_data_files(self, tmp_path):
+        got = self._orphans(tmp_path, ["fetch.py", "data.json", "notes.txt", "out.csv", "run.log"])
+        assert got == []  # none are viewable end-products
+
+    def test_presents_viewable_products(self, tmp_path):
+        got = self._orphans(tmp_path, ["report.html", "diagram.svg", "chart.png", "scratch.py"])
+        names = {p.rsplit("/", 1)[-1] for p in got}
+        assert names == {"report.html", "diagram.svg", "chart.png"}  # .py excluded
+
+    def test_skips_already_presented_and_screenshots(self, tmp_path):
+        got = self._orphans(
+            tmp_path,
+            ["report.html", "report.screenshot.png"],
+            already=["/mnt/user-data/outputs/report.html"],
+        )
+        assert got == []  # report.html already presented; screenshot is a sidecar