Skip to content

DeepTutor: Lifelong Personalized Tutoring. https://deeptutor.info/.

fix(reading): expose figure tool in lazy capability descriptor

Bingxi Zhao (Frank)6af0cc93607b
diff --git a/deeptutor/capabilities/reading/figure_view.py b/deeptutor/capabilities/reading/figure_view.py
index affd2185d..bbe1a0f63 100644
--- a/deeptutor/capabilities/reading/figure_view.py
+++ b/deeptutor/capabilities/reading/figure_view.py
@@ -34,8 +34,8 @@ class ViewFigureTool(_ReadingToolBase):
The chat turn only carries the images on the page the user has open, so any
other figure is invisible to a text model. This tool closes that gap by
resolving the named image on disk, asking a vision model about it, and
- returning the *text* — the only channel a tool result has. The pixels never
- travel; the answer does.
+ returning the *text* — the only channel a tool result has. The pixels go to
+ the selected vision model; only its answer enters the reading turn.
"""
name = "view_figure"
diff --git a/deeptutor/capabilities/reading/mode.py b/deeptutor/capabilities/reading/mode.py
index eac20e462..66c95d44d 100644
--- a/deeptutor/capabilities/reading/mode.py
+++ b/deeptutor/capabilities/reading/mode.py
@@ -4,7 +4,7 @@ There is no bespoke pipeline: the standard agentic chat loop IS the reader. This
capability only marks the turn and runs that pipeline. Everything specific to
reading is contributed by the loop capability
(:class:`deeptutor.capabilities.reading.capability.ReadingCapability`), which
-mounts the five reading tools, binds the open material server-side, injects the
+mounts the reading tools, binds the open material server-side, injects the
reading playbook and runs the deterministic locate pre-pass.
The split matters for a practical reason: the *mode* is what the user picks in
diff --git a/deeptutor/runtime/bootstrap/builtin_capabilities.py b/deeptutor/runtime/bootstrap/builtin_capabilities.py
index 1f78e5eba..3ace381ed 100644
--- a/deeptutor/runtime/bootstrap/builtin_capabilities.py
+++ b/deeptutor/runtime/bootstrap/builtin_capabilities.py
@@ -198,6 +198,7 @@ BUILTIN_CAPABILITY_SPECS: dict[str, BuiltinCapabilitySpec] = {
"material_outline",
"search_material",
"read_material",
+ "view_figure",
"reader_goto",
"reader_annotate",
"web_search",