Ecosystem Glossary¶

Plain-language definitions of every term a policy reader will encounter in an FCC audit bundle, brief, or discussion. Alphabetical order.

ADR (Architecture Decision Record)¶

A short Markdown document (usually 1-3 pages) that records a single architectural decision, the context, the alternatives considered, and the consequences. FCC ships five ADRs. Policy staff read ADRs to understand why a system is built a particular way. See docs/decisions/.

Archetype¶

A named pattern for a class of persona — e.g., Knowledge Finder, Synthesizer, Critic, Champion. The archetype atlas lists 37 ecosystem archetypes plus 26 research-backed variants. For a policy reader the archetype is a one-word summary of what the persona does.

Champion persona¶

An elevated persona that coordinates a team of base personas. Champions declare champion_of and orchestrates fields. Policy relevance: if a champion is high-risk, the team it orchestrates inherits extra scrutiny.

CLEAR+¶

FCC's seven-dimension benchmarking suite: Cost, Latency, Efficacy, Assurance, Reliability, Coverage, Explainability. CLEAR+ results appear in model cards and in the evidence graph as performance evidence.

Compliance Report¶

The JSON artifact produced by fcc compliance-audit. Contains a summary, findings, and pointers to evidence. See the Audit Primer.

Constellation-vertical¶

One of ten independent GitHub repos that provide vertical vocabularies on top of FCC: Ophiuchus (finance), Serpens (health), Libra (justice/legal), Crater (legal-records), Scutum (insurance), Norma (measurement/metrology), Pyxis (logistics), Vela (energy), Columba (messaging/communications), Caelum (atmosphere/environment). Each is versioned independently (currently v0.2.0).

Discernment Matrix¶

A 6-trait × 7-dimension rubric used in persona evaluation: Humility, Professional Background, Curiosity, Taste, Inclusivity, Responsibility rated on seven axes. Policy staff typically do not need to read raw matrix outputs, but they are referenced in evidence graphs.

EU AI Act (Regulation 2024/1689)¶

The European Union regulation on artificial intelligence. FCC packages 256+ machine-readable requirements derived from the Act. Personas are classified into risk categories from the Act's Article 5-7 scheme.

Evidence graph¶

A JSON-LD linked-data document connecting requirements, findings, and evidence items. Queryable. The most defensible artifact in legal review.

FCC (Find → Create → Critique)¶

The three-phase workflow that defines this framework. Also the name of the owning project.

Finding¶

A single item in the compliance report. Each finding has a status (compliant, partial, non_compliant, not_applicable), a requirement ID, evidence links, and optional remediation.

Governance category¶

The persona category whose members are concerned with policy, audit, oversight, ethics, and quality gates. Personas in this category are disproportionately likely to be high-risk under the AI Act.

Human-in-the-loop (HITL)¶

A deployment mode where human reviewers gate outputs before they leave the system. FCC supports HITL via the collaboration module.

LYRA¶

A sister project that provides a graph-backed UX layer for FCC ecosystems. Stable bridge to FCC as of v1.5.0.

Minimal / Limited / High / Unacceptable risk¶

The four EU AI Act risk categories. In FCC YAML, each persona carries risk_category with one of these values. A deployment's overall classification is the maximum of its personas.

Model card¶

A Markdown document per persona documenting intended use, out-of-scope uses, performance, and ethical considerations. Follows Mitchell et al. 2019.

NIST AI RMF¶

The US National Institute of Standards and Technology's AI Risk Management Framework (version 1.0, 2023). FCC maps 29 subcategories with bidirectional crosswalk to the EU AI Act.

Persona¶

A YAML file defining one role in an FCC system. Has ten R.I.S.C.E.A.R. fields plus metadata (category, risk_category, champion_of, etc.).

POLARIS¶

A sister project providing policy-ontology reasoning. Stable bridge to FCC as of v1.5.0.

R.I.S.C.E.A.R.¶

The ten-component persona specification: Role, Input, Style, Constraints, Expected Output, Archetype, Responsibilities, Role Skills, Role Collaborators, Role Adoption Checklist. Owned and defined by this project.

Remediation¶

A planned action to close a non-compliant or partial finding. Includes owner and due_date.

Risk heatmap¶

A visual summary of persona risk categories and open finding counts. Useful for briefings.

Scenario¶

A JSON file describing one FCC run: personas, inputs, workflow graph. FCC ships 33 scenarios.

Vertical pack¶

A bundled set of personas for a domain. Six vertical packs ship: healthcare (11), finance (11), legal (5), insurance (6), energy (6), government (6).

Workflow graph¶

A directed acyclic graph of workflow nodes. Seven canonical graphs ship, ranging from 5 to 84 nodes plus two solution-level EAIFC graphs.

Zachman cross-cut¶

An enterprise-architecture cross-reference matrix integrated into FCC v1.4.1+. Policy staff rarely need to read it; architects use it to verify coverage.