Glossary
Quick lookup for the abbreviations and TnsAI-specific terms that appear in the rest of the documentation. Linked terms go to the page that explains the concept in depth.
| Term | Definition |
|---|---|
| Action | A unit of work an agent can execute. Declared via @ActionSpec on a role method. |
| Agent | Long-lived orchestrator. The chat loop is LLM + registered tools, not a BDI interpreter. Agent has no belief/desire/intention getters. |
| AgentAdvisor | Planned interceptor SPI (beforeModel / afterModel / beforeTool / afterTool). Not on Maven Central 0.14.1. See TAN-5769 and AOSE. |
| AOSE | Agent-oriented software engineering. TnsAI’s contributor framing — roles, goals, organizations, protocols — mapped onto types in AOSE methodology. |
| BDI | Belief–Desire–Intention model types (Belief, Desire, Intention) plus the CognitiveModel.bdi() SPI. Not seeded on AgentBuilder (TNS-541). |
| Capability | A reusable contract (interface with default-method actions) a role can implement. |
| Channel | Adapter for an external messaging platform — Telegram, Slack, CLI, Email, etc. |
| FSM | Finite State Machine. The agent lifecycle (CREATED → STARTING → RUNNING → STOPPING → STOPPED + FAILED) is one. |
| GOAP | Goal-Oriented Action Planning. One of the planners under tnsai-intelligence. |
| HTN | Hierarchical Task Network — the other built-in planning style. |
| KB / Knowledge Base | Vector + BM25-indexed corpus that backs @KnowledgeSource / @Retrieval. |
| LAM | Language-Action Model — the pattern that maps LLM output to typed action calls. See LAM pattern. |
| MagenticFormation | Planned orchestrator + specialist Agent roster (WebSurfer, FileSurfer, Coder, Terminal). Not on Maven Central 0.14.1. See TAN-5773 and AOSE. |
| MCP | Model Context Protocol. Lets agents expose tools to (or consume tools from) MCP-compatible peers. |
| PlannerType | Planned public enum GOAP | UTILITY | HYBRID. Not on Maven Central 0.14.1. See TAN-5767 and AOSE. |
| RAG | Retrieval-Augmented Generation. |
| ReAct | Reasoning-and-Acting pattern: interleave LLM "thought" steps with tool calls. |
| ResolutionMode | Planned FOCUSED | CLOSED | OPEN agent selection. Not on Maven Central 0.14.1. See TAN-5768 and AOSE. |
| Role | Bundle of capabilities and actions an agent plays. |
| SCOP | A separate Java multi-agent framework — see scop-framework.netlify.app. TnsAI ships an integration bridge. |
| SPI | Service Provider Interface — Java's ServiceLoader discovery pattern, the framework's main extensibility seam. |
| ToT | Tree of Thoughts — branching reasoning pattern that explores multiple lines before committing. |
TnsAI as a Language-Action Model (LAM)
Industry writing calls this a Large Action Model: an LLM that does not stop at text, but selects and runs typed actions. In TnsAI that mapping is @ActionSpec methods plus registered tools. The chat loop is LLM + tools, not a BDI interpreter — see Roles.
Contracts (`@Contract`)
@Contract brings Design by Contract to actions: JEXL preconditions, postconditions, and invariants that the framework enforces around every action invocation. Preconditions reject bad input (including parameters an LLM hallucinated) before the method runs; postconditions catch incorrect results or corrupted state after it runs.