Skip to content

shadowhound_skills API Reference

Auto-generated Python API documentation.

Module: shadowhound_skills.skill_test_node

Functions

main()

No documentation available.


Module: shadowhound_skills.vision

Vision skills for ShadowHound using DIMOS Qwen VLM.

These skills leverage DIMOS's existing vision infrastructure: - Qwen VLM for scene understanding and object detection - Simple, synchronous API for easy testing and integration - Structured output for agent consumption

Class: SkillResult

Result from skill execution.

Class: VisionSkillBase

Base class for vision skills.

Methods

__init__(self, image_dir: Optional[Path], require_dimos: bool)

Initialize vision skill.

Args: image_dir: Directory to save captured images (default: /tmp/shadowhound/images) require_dimos: Whether this skill requires DIMOS vision APIs

Class: SnapshotSkill

Inherits: VisionSkillBase

Capture and save current camera frame. Does not require DIMOS.

Methods

__init__(self, image_dir: Optional[Path])

No documentation available.

execute(self, image: np.ndarray, **kwargs) -> SkillResult

Capture current frame and save with metadata.

Args: image: Current camera frame (numpy array or PIL Image) **kwargs: Optional metadata (pose, timestamp, etc.)

Returns: SkillResult with image path and metadata

Class: DescribeSceneSkill

Inherits: VisionSkillBase

Use VLM to describe what's in the image. Requires DIMOS.

Methods

__init__(self, image_dir: Optional[Path])

No documentation available.

execute(self, image: np.ndarray, query: Optional[str], save_image: bool, **kwargs) -> SkillResult

Get VLM description of the scene.

Args: image: Current camera frame query: Custom query (default: "Describe what you see in detail") save_image: Whether to save the image **kwargs: Additional parameters

Returns: SkillResult with VLM description

Class: LocateObjectSkill

Inherits: VisionSkillBase

Find specific object and return bounding box. Requires DIMOS.

Methods

__init__(self, image_dir: Optional[Path])

No documentation available.

execute(self, image: np.ndarray, object_name: str, save_image: bool, **kwargs) -> SkillResult

Locate specific object in the scene.

Args: image: Current camera frame object_name: Name of object to locate (e.g., "person", "chair") save_image: Whether to save the image **kwargs: Additional parameters

Returns: SkillResult with bounding box if found

Class: DetectObjectsSkill

Inherits: VisionSkillBase

Detect all prominent objects in the scene. Requires DIMOS.

Methods

__init__(self, image_dir: Optional[Path])

No documentation available.

execute(self, image: np.ndarray, save_image: bool, **kwargs) -> SkillResult

Detect all prominent objects in the scene.

Args: image: Current camera frame save_image: Whether to save the image **kwargs: Additional parameters

Returns: SkillResult with list of detected objects

Functions

get_skill(name: str) -> Optional[VisionSkillBase]

Get skill by name.

list_skills() -> list

List all available skills.


References