Glossary¶
Domain terms and definitions for the Animus project.
A¶
ADR — Architecture Decision Record. Documents a significant design choice: context, options, decision, tradeoffs. See docs/architecture/decisions/.
ADL — Arete Decision Log. A higher-level commitment scoped to a project milestone. ADLs live alongside ADRs.
C¶
Contracts — Canonical JSON schemas that define data structures across Animus subsystems. 20+ schemas in packages/contracts/.
Core — The personal AI exocortex package (animus). Handles memory, CLI, integrations, and the cognitive layer.
Crucible — The universal transformation framework for navigating change. Phase detection, failure taxonomy, active/receptive polarity.
E¶
Exocortex — An external cognitive system that augments biological intelligence. Persistent memory, task tracking, preference learning across sessions and devices.
F¶
Forge — Multi-agent workflow orchestration engine (animus_forge). YAML-defined pipelines with budget controls and checkpoint/resume.
J¶
Jidoka — Stop and fix before propagating errors. One of the TPS principles applied to software.
K¶
Kaizen — Continuous improvement philosophy. Applied to Animus via the reflection loop and self-improvement pipeline.
Kernel — Autonomous builder engine (animus_kernel). Budget, executor, sandbox, safety, multi-agent supervision.
M¶
Mind-class — The target architecture for Animus v2.1. A persistent, self-improving personal intelligence with memory, planning, and autonomous execution.
P¶
Poka-yoke — Error prevention built into the system. Used in Forge's safety checks and config validation.
Q¶
Quorum — Agent coordination protocol (convergent). Decentralized multi-agent coherence via shared intent graph.
T¶
TPS — Toyota Production System. Applied to software via cost visibility, waste elimination, and quality gates.
Truth Baseline — Automated validation that documented claims match codebase reality. Run via scripts/truth-baseline.py.
Types — Shared schema definitions (animus_types). Install first as a sibling dependency.
V¶
v2.1 — The canonical executable baseline for Mind-class architecture. 8 technical planes, 22 JSON schemas, PostgreSQL durable core.
See Also¶
- Concepts — Core mental models
- Architecture → Overview — System architecture