Please enable JavaScript.
Coggle requires JavaScript to display documents.
Taxonomy of AgentOps (Observability) Artifacts - Coggle Diagram
Taxonomy of AgentOps (Observability) Artifacts
Agent Input Artifact
Agent Registry
Basic
Agent ID
Agent Version
Agent Design Type
Agent Role
Agent Architecture
Agent Guardrails
Additional
Prompt
Input Data
LLM Model
ToolKit
Prompt
Prompt Registry
Basic
Prompt ID
Prompt Version
Prompt Template
Prompt Optimisation
Additional
Prompt Model
Prompt Variables
Prompt Test Result
Prompt Template Component
User Goal
Input Query
Few-shot Example
Context Doc
Instruction
Output Structure
Prompt Optimisation Type
Zero-Shot Learning
Plan-and-Solve Prompting
Self-Consistency
Iterative Prompting
Chain-of-Thought(CoT) Prompting
Few-Shot Learning
Agent Input Data
Input Data Source
RAG(Query, Retrieval Doc, Retrieval Keyword, Embedding vectors )
In-Context Learning(Task Description, Context Example, Prompt)
Input Data Type
Text-based
Table
Images
Video
Voice
Agent Design
Agent DesignType
Self Reflection Agent
Self-Discover
Tree Search
Basic Reflection
Planning Agent
LLMCompiler: Plan -> Action in parallel->joint
Plan and Solve: Plan -> Task List->Replan
Re-Act:
Default (Think -> Execute)
FixedTask (Think -> Prioritize -> Execute)
DynamicTask (Think -> Prioritize -> Execute-> Add Task -> Execute)
Agent Role
Worker
Creater
Coordinator
Agent Architecture
Single Agent
Multi-Agent
Guardrails
Guardrails Rules
-Uniform/Priority-enabled strategy/Context-dependent/Negotiability
Guardrails Targets Step
-Goal/Context/Memory/Reasoning/Plans/Actions/Tools/Intermediate Results/ Final Results
Guardrails Actions
-Block/Filter/Flag/Fall back/Human intervention/Defer Isolate ...
Agent Evaluation and Test
Benchmark Evaluation
Evaluation Dataset
Evaluation Metrics
Evaluation Criteria
Agent Evaluation
Final Response Evaluation
Single Step Evaluation
Trajectory Evaluation
Evaluation Registry
Evaluation Version
Evaluation Test Cases/Script
Evaluation Input
Evaluation Output
Evaluation Method
Evaluation Result
Feedback
Direct(User) Feedback
Indirect Feedback
User behaviour
Reflection Result
Agent Monitoring
Monitering Metrics
Session-level Metrics
Event-level Metrics
Agent Error Issues
Agent Execution
Action Event
Action ID
Responding Task
LLM Call
Tool Call
Tool Event
Tool Name
Responding Task
Responding Action
Tool Executed Result
LLM Call Event
Input
Goal/Prompt/Context/External Doc
Memory
Short-term Memory(recent interactions, intermediate outcomes)
Long-term Memory(Knowledge database, past execution )
Entity Memory (people, places, concepts encountered during tasks)
LLM Model
Model Name/LLM Function Call ID
LLM Call Cost
Prompt Token
Completion Tokens
Planning Event
Planning ID
Planning Agent Architecture
Task List(Entire Task/Subtasks)
Agent Role
Agent Cooperation
Agent Output
Intermediate Output
LLM Call/Tool Call/Subtask Execution Result
Final Response
Answer Quality/Output Guardrail/...
Agent Workflow Chain
Interaction data
Decision Process
Feedback