How iambmd/ttskill works
TTSkill is a table-tennis video coaching product. Players submit a clip or YouTube link through one of four tracks: auto-cut rallies, get a technique breakdown, find comparable pros, or ask TT Coach to coordinate several tools. The frontend job APIs reach FastAPI, which validates input, creates background jobs, and runs the relevant pipeline; results and generated videos are shown on job pages. The graph preserves equipment recommendations and coaching-style analysis as separate capabilities. Some internal wiring is unsampled, so uncertain links are omitted; documented workflow links cite the README.
Main entry points: pipelines.py Pipeline orchestration · main.py FastAPI routes · pipeline.py Technique pipeline · jobs.py Job records
Components 23
Player experience
- Track pages
frontend/app/autocut/page.tsx - Job API routes
frontend/app/api/jobs/autocut/route.ts - Job results
frontend/app/jobs/[id]/page.tsx
Jobs and orchestration
- FastAPI routes
backend/app/main.py - Input validation
backend/app/validate.py - Job records
backend/app/jobs.py - Pipeline orchestration
backend/app/pipelines.py - Video download
backend/app/download.py
Video analysis
- Rally detection
backend/app/rally.py - Rally filtering
backend/app/floor_bounce.py - Clip editing
backend/app/cut.py - Technique pipeline
backend/technique/pipeline.py - Pose analysis
backend/technique/pose.py - Stroke analysis
backend/technique/strokes.py - Media processing
backend/technique/media.py - Breakdown renderer
backend/technique/render.py
Coaching and matching
- TT Coach agent
backend/agent/coach.py - Coach tools
backend/agent/tools.py - Pro matching
backend/players/match.py - Body profiles
backend/players/body.py
Equipment and style
- Rubber recommendations
backend/equipment/recommend.py - Coaching style
backend/technique/style.py
Other components
- Player
Connections 25
- Player → Track pages: chooses track
- Track pages → Job API routes: submits job
- Job API routes → FastAPI routes: forwards request
- FastAPI routes → Input validation: validates input
backend/app/main.py - FastAPI routes → Job records: creates job
backend/app/main.py - FastAPI routes → Pipeline orchestration: starts pipeline
backend/app/main.py - Pipeline orchestration → Job records: updates status
backend/app/pipelines.py - FastAPI routes → Video download: fetches video
backend/app/main.py - Pipeline orchestration → Rally detection: detects rallies
backend/app/pipelines.py - Pipeline orchestration → Rally filtering: filters rallies
backend/app/pipelines.py - Pipeline orchestration → Clip editing: cuts clips
backend/app/pipelines.py - Rally filtering → Rally detection: uses candidates
backend/app/floor_bounce.py - Pipeline orchestration → Technique pipeline: runs breakdown
backend/app/pipelines.py - Technique pipeline → Media processing: makes proxy
backend/technique/pipeline.py - Technique pipeline → Pose analysis: extracts pose
backend/technique/pipeline.py - Technique pipeline → Stroke analysis: plans stroke
backend/technique/pipeline.py - Technique pipeline → Breakdown renderer: renders video
backend/technique/pipeline.py - FastAPI routes → TT Coach agent: starts coach
backend/app/main.py - TT Coach agent → Coach tools: dispatches tools
backend/agent/coach.py - Coach tools → Pipeline orchestration: calls shared work
backend/agent/tools.py - Pipeline orchestration → Pro matching: ranks pros
backend/app/pipelines.py - Pro matching → Body profiles: compares profiles
backend/players/match.py - FastAPI routes → Rubber recommendations: recommends rubbers
backend/app/main.py - Breakdown renderer → Coaching style: uses layout style
backend/technique/render.py - Job records → Job results: serves results