How nader20-5/gillo-ai-services works
Taskove (Gillo) is a FastAPI AI service for an Agile platform. Its main workflow serves sprint analysis and task estimates through API routers, using task/sprint read models, feature engineering, and a versioned model registry with rule fallbacks; sprint analysis can also produce risk suggestions. Separate API branches support suggestion management, AI drafts, and chat. Background jobs repeat risk scans and maintain suggestion/profile state. The README also documents Model 1 estimation and Model 2 carry-over risk capabilities. Edges are limited to relationships visible in excerpts, the SOURCE INDEX, or the README; unsampled subsystems are represented without speculative wiring.
Main entry points: sprints.py Sprint analysis · sprint_jobs.py Sprint background jobs · main.py FastAPI entrypoint · database.py MongoDB access
Components 22
API and access
- FastAPI entrypoint
app/main.py - JWT access control
app/middleware/auth.py
Sprint intelligence
- Sprint analysis
app/routers/sprints.py - Estimation service
app/services/estimation_service.py - Feature engineering
app/ml/features.py - Model registry
app/ml/registry.py - Agile read models
app/models/read_models.py - Suggestion inbox
app/routers/suggestions.py - Offline features
benchmark_v2/features_v2.py - Benchmark replay
benchmark_v2/replay.py
AI interactions
- Draft generation
app/routers/drafts.py - AI chat
app/routers/chat.py - Chat tools
app/services/chat_tools.py - LLM providers
app/services/llm_provider.py - LLM service
app/services/llm.py - Model 3 metrics
app/evaluation/model3_metrics.py
State and operations
- MongoDB access
app/database.py - Sprint background jobs
app/jobs/sprint_jobs.py - Job scheduler
app/jobs/scheduler.py - Event consumer
app/events/consumer.py - Privacy controls
app/utils/privacy.py
Other components
- Agile user
Connections 30
- Agile user → FastAPI entrypoint: submits requests
app/main.py - FastAPI entrypoint → Sprint analysis: routes requests
app/main.py - FastAPI entrypoint → Suggestion inbox: routes requests
app/main.py - FastAPI entrypoint → Draft generation: routes requests
app/main.py - FastAPI entrypoint → AI chat: routes requests
app/main.py - FastAPI entrypoint → Job scheduler: starts jobs
app/main.py - Sprint analysis → MongoDB access: reads and writes
app/routers/sprints.py - Sprint analysis → Estimation service: computes estimates
app/routers/sprints.py - Sprint analysis → Feature engineering: builds features
app/routers/sprints.py - Sprint analysis → Model registry: scores tasks
app/routers/sprints.py - Sprint analysis → Agile read models: uses snapshots
app/routers/sprints.py - Sprint analysis → Suggestion inbox: creates warnings
- Suggestion inbox → MongoDB access: reads and writes
app/routers/suggestions.py - Draft generation → MongoDB access: reads and writes
app/routers/drafts.py - Draft generation → LLM service: requests generation
app/routers/drafts.py - LLM service → LLM providers: delegates generation
app/services/llm.py - AI chat → Chat tools: uses tools
app/routers/chat.py - AI chat → LLM providers: requests generation
app/routers/chat.py - AI chat → MongoDB access: reads and writes
app/routers/chat.py - Job scheduler → Sprint background jobs: schedules jobs
app/jobs/scheduler.py - Sprint background jobs → MongoDB access: reads and writes
app/jobs/sprint_jobs.py - Sprint background jobs → Agile read models: builds snapshots
app/jobs/sprint_jobs.py - Sprint background jobs → Model registry: scores sprints
app/jobs/sprint_jobs.py - Sprint background jobs → Feature engineering: computes features
app/jobs/sprint_jobs.py - Sprint background jobs → Estimation service: refreshes ratios
app/jobs/sprint_jobs.py - Event consumer → Sprint background jobs: invokes job logic
app/events/consumer.py - Event consumer → Privacy controls: uses controls
app/events/consumer.py - Model registry → Feature engineering: validates contract
app/ml/registry.py - Model registry → Offline features: uses feature definitions
app/ml/registry.py - Benchmark replay → Offline features: uses features
benchmark_v2/scripts/run_model2_final_benchmarks.py