CoffeeBreakAI Glossary
A unified reference of all terms used throughout the CoffeeBreakAI platform and documentation. This glossary includes system concepts, agent mechanics, HITL workflows, workspace settings, integrations, and advanced developer-level terms.
🧩 Core System Concepts
Task
A discrete unit of work submitted to CoffeeBreakAI. Examples: implementing a feature, refining a pull request, decomposing an epic, or applying review feedback.
Agent
A specialized AI worker responsible for executing a specific category of task (coding, planning, analysis, refactoring, decomposition, etc.).
Agent Registry
The component that tracks all available agents, their capabilities, and how to invoke them.
Task Router
The orchestration engine that chooses which agent or workflow should handle a task based on metadata, capability hints, task type, and workspace rules.
TaskExecution
The persisted record of a task’s entire lifecycle, including agent output, retries, review results, timestamps, metadata, and routing decisions.
Trigger Source
The origin of a task, such as an initial user request, dependency completion, retry event, or reviewer feedback.
Dependency Graph
A data structure describing which tasks depend on others. Tasks may be blocked until prerequisite tasks complete.
Priority / Task Priority
A ranking used to influence the order in which tasks are processed.
🏛️ Human-in-the-Loop (HITL)
Human Review (HITL)
A stage where CoffeeBreakAI pauses a task and waits for human approval, rejection, or feedback.
Reviewer
A human authorized to review task output.
Reviewer Groups
Workspace-defined groups of reviewers assigned to specific task types or workflows.
Review
A review action performed on a task: approve, reject, or comment.
Reviewer Feedback
The comments (human or AI-generated) that CoffeeBreakAI uses to refine work or trigger follow-up tasks.
Refinement Loop
The iterative process where CoffeeBreakAI updates a pull request or task output based on reviewer requests.
📐 Quality, Evaluation & Retries
Confidence Score
A numeric estimate (0–1) of how confident CoffeeBreakAI is that an agent’s output is correct without human review.
Confidence Threshold
The minimum required confidence score before CoffeeBreakAI can proceed autonomously.
Retry
A re-attempt triggered when CoffeeBreakAI detects a recoverable issue (formatting, test failures, transient external errors, etc.).
Retry Rules
Workspace-level policies defining retry attempts, exponential backoff, classification of recoverable vs. non-recoverable errors, and escalation behavior.
🔌 Integrations & External Systems
Integration
A connection between CoffeeBreakAI and an external system (GitHub, Jira, Azure DevOps, etc.).
Git Provider
Any supported Git platform (GitHub, GitLab, Bitbucket, etc.).
Pull Request (PR)
A code change that CoffeeBreakAI can generate, update, refine, or review.
External Work Item
Issues or tasks originating from Jira, GitHub Issues, Azure Boards, etc., and linked into CoffeeBreakAI.
📚 Knowledge, Context & Metadata
Capability Hint
Metadata describing skills, domains, or patterns relevant to a task. Used for agent routing and refinement.
Agent Metadata
Describes what an agent does, which capabilities it advertises, and how it participates in routing and refinement loops.
Shared Context Store
Workspace-level documents and knowledge that CoffeeBreakAI can reference to maintain consistency with standards and architecture.
Documentation Indexing
The ingestion process that allows CoffeeBreakAI to understand and use your documentation during planning, coding, and refinement.
🧪 Advanced and Developer-Level Concepts
Custom Agent
A user-defined agent with custom instructions, behaviors, and capabilities.
Strategy
A defined behavioral pattern CoffeeBreakAI uses to perform tasks or improve decisions.
Strategy Evolution
CoffeeBreakAI’s internal mechanism for improving strategies using historical task outcomes, review feedback, and success patterns.
🗂️ Workspace Concepts
Workspace
A logical container that includes tasks, reviewers, settings, documentation, integrations, and agent configurations.
Reviewer Permissions
Rules controlling who can approve, reject, or resume paused tasks.
📄 Documentation & Change Management
Changelog
A record of externally visible updates or changes to CoffeeBreakAI.
🎉 Need Help?
See: