README.md

metadata

title: skillsculptor
app_file: app.py
sdk: gradio
sdk_version: 5.44.1

SkillSculptor

Multi-agent workforce planning assistant that analyzes skills, identifies project staffing gaps, and suggests upskilling paths. A coordinator agent orchestrates specialized worker agents (skill matching, gap coverage, upskilling, reporting, and guardrails) and serves responses via a Gradio chat UI with optional short-term or long-term memory.

Features

• Multi-agent orchestration via a coordinator
• Worker agents for skill matching, gap coverage, upskilling, reporting, and guardrails
• Gradio chat interface with streaming and memory mode toggle
• Sample data and generated reports saved locally

Project structure

app.py                      # Gradio app + chat handler
worker_agents/              # Coordinator + worker agents
  coordinator.py
  skill_matching.py
  gap_coverage.py
  upskill.py
  reporter.py
  guardian.py
tools/                      # Tooling used by agents
ai_clients/                 # Model client(s), e.g., Gemini
utils/                      # Guardrails and utilities
data/                       # Sample CSV datasets (employees, projects, trends, upskill)
reports/                    # Generated reports (gap report, suggestions, violations)
requirements.txt            # Python dependencies

Prerequisites

• Python 3.9–3.11 recommended
• API keys
  • Google Gemini: set GOOGLE_API_KEY (free; app can run solely on Gemini)
  • OpenAI (optional): set OPENAI_API_KEY to enable OpenAI tracing logs
  • Without OPENAI_API_KEY, the app still runs; OpenAI tracing is disabled

Setup (Windows PowerShell)

# From the repo root
python -m venv .venv
.\.venv\Scripts\Activate.ps1
python -m pip install --upgrade pip
pip install -r requirements.txt

# Optionally store keys in a .env file at the project root
# (python-dotenv is included; many agent clients honor .env)
# Create .env with lines like:
# GOOGLE_API_KEY=...
# OPENAI_API_KEY=sk-...   # optional, only needed for OpenAI tracing logs

Run locally

# From the activated virtual environment
python app.py

• The app launches a Gradio UI (auto-opens browser). If not, open the printed http://127.0.0.1:7860 URL.
• In the UI, choose memory mode (Short-term vs Long-term) and start chatting about staffing, skills, or gaps.

Data and outputs

• Input data: CSVs in sample_data/
• Generated reports: Markdown/text in reports/

Tracing and observability

• The app emits traces (see app.py usage of trace). When OPENAI_API_KEY is set, traces are sent to OpenAI tracing logs for observability and debugging.
• If you omit OPENAI_API_KEY, the app still runs (especially with Gemini) and incurs no OpenAI cost; OpenAI tracing is simply disabled.

Notes

• If you use Gemini, ensure ai_clients/gemini_client.py is configured and GOOGLE_API_KEY is set.
• To extend behavior, add/modify worker agents in worker_agents/ and tools in tools/.
• Cost note: Running with Gemini only is free; OpenAI key is used for tracing, not required for core functionality.