Structure

Layers are inferred from where files sit in the tree, not from reading the code. A project that names its directories unconventionally will read oddly here — open the file browser to check anything the diagram implies.

Languages

Share of indexed source by file size. Binary and vendored files are excluded.

Dependencies

Declared in the repository’s manifests at the indexed commit. A declared package is not proof it is used, and runtime dependencies are listed first.

Feature verification

• Flask REST backendVerified

  REST API written in Python using Flask

  Claimed on readmehigh confidence

  • backend/backend/app.py:18 — Flask app with /video and /process_audio routes
• Frontend-to-robot-server proxy routesVerified

  Frontend makes requests to robot control backend based on keyboard and mouse inputs

  Claimed on readmehigh confidence

  • spotter/pages/api/robot/[move].ts:28 — Next.js API route proxies validated move commands to http://127.0.0.1:5000/ (robot-control-server) move/turn/sit/stand endpoints
• Next.js frontend with robot control and camera/data feedVerified

  Frontend that enables easy robot control and shows the robot's camera and data feed

  Claimed on readmehigh confidence

  • spotter/components/controls.tsx:12 — Controls component sends WASD/keyboard and button inputs to /api/robot/[move]
  • spotter/components/webcam.tsx:51 — WebCam component renders the backend's live MJPEG video feed at localhost:5001/video
  • spotter/components/transcription-table.tsx:114 — TranscriptionTable fetches /api/audio and displays emotion/urgency/summary data feed
• Person detection via OpenCV in live video feedVerified

  Uses OpenCV for ML models to assess disaster zones

  Claimed on Devposthigh confidence

  • backend/backend/app.py:22 — generate_frames() uses cv2 HOGDescriptor people detector to draw bounding boxes on webcam frames
  • backend/backend/app.py:42 — /video Flask route streams the annotated MJPEG feed
• Speech-to-text transcription (OpenAI Whisper)Verified

  Understands natural language using AI, multimodal AI processing backend

  Claimed on readmehigh confidence

  • backend/backend/app.py:147 — records 5s of mic audio and calls openai.Audio.transcribe('whisper-1', ...) to get transcript text
• SPOT robot motor control server (Boston Dynamics SDK)Verified

  Custom control server that connects directly to SPOT and controls its motors; robot-control-server REST API using Boston Dynamics SDK

  Claimed on readmehigh confidence

  • robot-control-server/python/examples/wasd_server/app.py:3 — imports bosdyn.client, authenticates and powers on the robot, exposes Flask routes like /move/forward, /sit, /stand that call WasdInterface methods
  • robot-control-server/python/examples/wasd_server/wasd.py:1 — WasdInterface implementation referenced by app.py for movement/estop/power control
• GPT-4 conversational response generationCode-supported

  Uses GPT-4 for AI processing (README built-with)

  Claimed on readmemedium confidence

  • backend/backend/app.py:157 — Calls openai.Completion.create with engine 'text-davinci-002', a GPT-3 completion model, not GPT-4, contradicting the README claim
• Hume AI emotion/prosody analysis of audioCode-supported

  Uses Hume for AI/transcription and emotion understanding

  Claimed on Devpostmedium confidence

  • backend/backend/app.py:47 — postAudioHume() submits recorded audio to HumeBatchClient with ProsodyConfig and parses top emotions
  • backend/backend/app.py:20 — HUME_API_KEY is hardcoded empty string, so the call would fail without a real key at runtime; wrapped in try/except that falls back to a hardcoded stub emotion
• Language detection and response in the human's languageCode-supported

  Noted ability to detect a language and adjust output as such

  Claimed on Devpostmedium confidence

  • backend/backend/app.py:151 — GPT prompt instructs the model to 'reply in the language of the human', relying on the LLM to infer language from transcript rather than explicit language-detection code
  • backend/backend/app.py:161 — gTTS text-to-speech call is hardcoded to lang='en', which contradicts multilingual output despite the prompt instruction
• Urgency level and paramedic summary generationCode-supported

  Generates actionable insights/summary for first responders

  Claimed on Devpostmedium confidence

  • backend/backend/app.py:175 — urgency_level is hardcoded to 8 rather than computed from the emotion/text data
  • backend/backend/app.py:178 — summary_prompt asks GPT to synthesize an actionable paramedic summary from transcript, emotions, and urgency level
• Autonomous/independent disaster-zone navigation and survivor locationClaimed only

  The SPOT robot has been enabled to be an independent rescue machine ... traverse and navigate disaster environments completely independently to locate survivors

  Claimed on Devposthigh confidence
• Bark text-to-speechClaimed only

  Uses Bark (suno-ai) for speech synthesis

  Claimed on readmehigh confidence
• Facial emotion detection via HumeClaimed only

  Understands human emotion via AI (implies facial + audio)

  Claimed on Devpostmedium confidence
• Geolocation data transmission to rescue teamsClaimed only

  Images, voice, and geolocation data is being sent in real time to rescue teams

  Claimed on Devpostmedium confidence

An AI agent derived these features from the project’s Devpost page and readme, then searched the code for each one. Verified features are backed by cited code; claimed-only features had no supporting code, which is not by itself proof a feature is missing.