Glossary
This page lists common terms in Codeer.ai and their explanations.
A
Agent
AI assistant. You can create custom Agents in Codeer.ai, configuring their personality, expertise, and knowledge sources.
API Key
Authentication credentials for programmatic access to Codeer.ai services. Should be kept secure and not exposed.
C
Channel
Agent's publishing channel. You can provide Agents to users through different channels (Web, API, Slack).
Connection
Configuration for connecting external data sources (like GitHub, Notion) to Codeer.ai.
Context
Background information AI references when answering questions, including conversation history and connected data sources.
Copilot
Codeer.ai's built-in floating AI assistant that can help you configure and manage Agents, with one-click suggestion application functionality.
E
Embedding
Technology that converts text into numerical vectors, used for semantic search and similarity comparison.
Evaluation / Test Suite
Feature for recording "how the Agent should answer." You can define input questions and expected standards, run tests after each Agent update to ensure quality doesn't regress.
I
Instruction
Instructions that define the Agent's role, behavior, and rules. This is key to configuring the Agent's personality.
L
LLM (Large Language Model)
Core technology powering AI assistants, such as GPT-4, Claude, etc.
O
OAuth
Authorization protocol used to securely connect external services (like GitHub, Google).
Organization
Top-level management unit in Codeer.ai, typically representing a company or team.
P
Playground
Main interface for conversing with Agents in Codeer.ai.
Prompt
Instructions or questions given to AI. Instruction defines Agent behavior; user prompt is the question input by users.
R
RAG (Retrieval-Augmented Generation)
AI technology that combines data retrieval and text generation, allowing AI to answer questions based on your data.
Repository
Place where code is stored, typically referring to projects on GitHub.
S
Sync
Process of updating connected data sources to Codeer.ai.
T
Temperature
Parameter controlling AI answer creativity. Lower temperature produces more consistent answers; higher temperature produces more diverse answers.
Test Suite
Token
Basic unit for AI models to process text. Roughly, 1 English word equals 1-2 tokens, 1 Chinese character equals 1-2 tokens.
V
Validator
Rules defining "what makes a good answer." When running tests, Validator scores Agent answers based on your set standards.
W
Workspace
Project or team unit under an organization. Each workspace has its own Agents, data connections, and members.
Related Resources
- Getting Started - Onboarding tutorial
- FAQ - Frequently Asked Questions