Skip to content

Frequently Asked Questions

Common questions and answers for newcomers to the FCC Agent Team Framework.

General Questions

What does FCC stand for?

FCC stands for Find, Create, Critique -- the three phases of the core workflow cycle. "Find" is about discovering and gathering information, "Create" is about synthesizing and building, and "Critique" is about reviewing and improving.

Do I need an AI API key to use FCC?

No. FCC includes a "mock" simulation mode that works entirely offline with no API keys. Mock mode is deterministic and ideal for learning, testing, and reproducibility. You only need an API key (Anthropic or OpenAI) if you want to run AI-powered simulations.

What programming language is FCC written in?

FCC is written in Python (3.10+). The web frontend uses TypeScript with React 18 and D3.js. Streamlit apps use Python.

How is FCC different from LangChain or AutoGen?

FCC is a persona-driven workflow framework, not a general-purpose AI orchestration library. Key differences:

  • Persona-first: Every agent has a full R.I.S.C.E.A.R. specification with 10 components, not just a system prompt.
  • Structured workflows: Uses graph-based workflows (7 variants) rather than free-form chains.
  • Governance built-in: 3-tier constitutions, quality gates, and audit trails are core features.
  • Knowledge-aware: Integrated knowledge graphs, semantic search, and RAG pipeline.
  • No vendor lock-in: Works with Anthropic, OpenAI, or fully offline in mock mode.

Is FCC open source?

Yes. FCC is released under the MIT License by INFORMATION COLLECTIVE, LLC.

Installation Questions

What Python version do I need?

Python 3.10 or higher. Check with:

python --version

How do I install FCC?

The recommended way is:

git clone https://github.com/rollingthunderfourtytwo-afk/l2_fcc_agent_team_ext.git
cd l2_fcc_agent_team_ext
make venv && source .venv/bin/activate
make install-dev

See GETTING_STARTED.md for all installation options.

Can I install FCC on Windows?

Yes. Use the standard Python installation process:

python -m venv .venv
.venv\Scripts\activate
pip install -e .

Note: make commands require Make for Windows (available via Chocolatey or MSYS2) or you can run the equivalent pip commands directly.

What are the optional dependencies for?

Extra Purpose
notebooks Jupyter notebooks (jupyter, matplotlib)
streamlit Streamlit apps (streamlit, plotly)
search Semantic search (sentence-transformers)
knowledge KG export (rdflib)
observability OpenTelemetry integration
dev Testing and linting (pytest, ruff)

Install specific extras with: pip install -e ".[notebooks,streamlit]"

Concept Questions

What is a persona?

A persona is a defined AI "expert" with a specific role, style, skills, and behavioral profile. Think of it as a detailed job description for an AI agent. Each persona has a R.I.S.C.E.A.R. specification that defines exactly how it should behave. FCC includes 102 built-in personas.

See the Glossary for a more detailed definition.

What is R.I.S.C.E.A.R.?

R.I.S.C.E.A.R. is the 10-component specification that defines every persona: Role, Input, Style, Constraints, Expected Output, Archetype, Responsibilities, Role Skills, Role Collaborators, and Role Adoption Checklist.

See SAMPLE_PROMPTS.md prompt B2 for a code example that prints all 10 components.

What is a simulation trace?

A trace is the complete record of a simulation run. It captures every step that was executed, including which persona acted, what phase it was in, and a summary of the output. Traces enable reproducibility -- you can replay a trace to verify results.

What are the 20 persona categories?

The categories are: app_development, champions, core, data_engineering, devops, docs_as_code, governance_compliance, integration_specialists, jv_collaboration, jv_governance, knowledge_graph, local_first_ai, ml_lifecycle, ml_models, open_science, privacy, protocol_engineering, responsible_ai, stakeholder_hub, and ux_visualization.

How many tests does FCC have?

FCC has 12,100+ tests with 99%+ code coverage. Run them with make test.

Usage Questions

How do I find a specific persona?

from fcc.personas.registry import PersonaRegistry

registry = PersonaRegistry.from_package_data()

# By ID
persona = registry.get("research_catalyst")

# By category
ml_personas = registry.by_category("ml_lifecycle")

# All champions
champions = registry.champions()

How do I run a simulation?

from fcc.personas.registry import PersonaRegistry
from fcc.simulation.engine import SimulationEngine
from fcc.scenarios.loader import ScenarioLoader

registry = PersonaRegistry.from_package_data()
scenarios = ScenarioLoader.from_package_data()
scenario = scenarios.get("basic_fcc_cycle")

engine = SimulationEngine(registry=registry, mode="mock")
trace = engine.run(scenario)

See SAMPLE_PROMPTS.md prompt B3 for a complete example.

How do I create a custom persona?

Write a YAML file following the persona schema, then load it into the registry. See Notebook 12_custom_persona_design.ipynb for a guided tutorial, or the Student Workbook Exercise 4.1.

Can I use FCC with my own data?

Yes. The RAG pipeline lets you ingest your own documents, and the knowledge graph lets you model your own domain concepts. See the for-scientists section for research-oriented examples.

How do I contribute to FCC?

Check the repository's GitHub Issues for open tasks. Follow the conventions documented in CLAUDE.md: use dataclasses (not Pydantic), click (not argparse), and keep source code under src/fcc/.

Troubleshooting

"ModuleNotFoundError: No module named 'fcc'"

Make sure your virtual environment is activated:

source .venv/bin/activate    # Linux/macOS
.venv\Scripts\activate       # Windows

Then verify FCC is installed:

pip show fcc

Tests are failing

Run with verbose output to see which tests fail:

pytest tests/ -v --tb=short

Common causes: - Wrong Python version (need 3.10+) - Missing optional dependencies (install with pip install -e ".[dev]")

Streamlit app shows "ImportError"

Install Streamlit dependencies:

pip install -e ".[streamlit]"

Frontend shows blank page

Make sure you installed Node.js dependencies:

cd frontend
npm install
npm run dev

Check that the development server is running on http://localhost:5173.

Where to Get Help