-
Changes before error encountered
copilot-swe-agent[bot] +1: IngSeb0
Agent-Logs-Url: https://github.com/IngSeb0/ReliefLensAI/sessions/a55020a9-6a8e-42a4-ac52-7612f271b585
Co-authored-by: IngSeb0 <141876214+IngSeb0@users.noreply.github.com> -
refactor: extract default location fallback to constant in triage_agent
copilot-swe-agent[bot] +1: IngSeb0
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Co-authored-by: IngSeb0 <141876214+IngSeb0@users.noreply.github.com> -
feat: build complete ReliefLensAI backend
copilot-swe-agent[bot] +1: IngSeb0
- FastAPI app with 5 routers: /crisis-room, /reports, /incidents, /amd, /demo
- 8 agents: intake → transcription → vision → normalization → dedup → triage → resource → dispatch
- 11 skills: transcribe_audio, caption_image, extract_location, normalize_signal,
detect_duplicates, classify_priority, recommend_resources, generate_dispatch_message,
calculate_confidence, fetch_amd_metrics, export_incident_report
- vLLM client wrapper with full demo_mode (no hardware required)
- JSON file-backed storage with in-memory cache
- End-to-end Pipeline orchestrating all agents
- 29-report Santa Ana flood scenario (text/audio/image/CSV, duplicates, P0-P3)
- 30 tests covering schemas, skills, and API endpoints
- Dockerfile, setup.sh, run_backend.sh scripts
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Co-authored-by: IngSeb0 <141876214+IngSeb0@users.noreply.github.com> -
fix: use resolve().parents for robust scenario path
copilot-swe-agent[bot] +1: IngSeb0
Avoids fragile relative parent traversal in demo route.
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Co-authored-by: IngSeb0 <141876214+IngSeb0@users.noreply.github.com> -
feat: build complete ReliefLensAI backend
copilot-swe-agent[bot] +1: IngSeb0
Full multimodal disaster triage backend including:
- FastAPI app with CORS, lifespan, health check
- Pydantic v2 schemas: report, signal, incident, resource, dispatch, amd, crisis_room
- 11 skills: transcribe_audio, caption_image, extract_location, normalize_signal,
detect_duplicates, classify_priority, recommend_resources, generate_dispatch_message,
calculate_confidence, fetch_amd_metrics, export_incident_report
- 8 agents: intake, transcription, vision, normalization, dedup, triage, resource, dispatch
- Services: vLLM client (demo+real mode), storage (file-backed JSON cache), pipeline
- API routes: /crisis-room, /reports, /incidents, /amd, /demo
- Demo data: 29-report Santa Ana flood scenario (text/audio/image/csv)
- Tests: 30 passing across schemas, skills, and API
- Docker, .env.example, setup/run scripts
- Demo mode enabled by default (no external dependencies required)
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Co-authored-by: IngSeb0 <141876214+In…
