This glossary defines the technical terms, internal jargon, and abbreviations used within the oh-my-openagent codebase. It serves as a reference for onboarding engineers to understand the implementation details and data flows associated with these concepts.
ulw)A high-level orchestration mode that engages the full suite of specialized agents to work autonomously on a task until completion. It bypasses manual intervention by leveraging the Ralph Loop and Boulder mechanisms to ensure progress.
ultrawork or ulw commands, which activate the Sisyphus orchestrator in a continuous loop README.md109IntentGate -> Sisyphus -> Specialized Subagents -> Task Completion.
Sources: README.md109A term used to describe agents (specifically Sisyphus) that are programmed to be "self-disciplined"—meaning they strictly follow verification loops, check their own work, and do not stop until the defined goal is met README.md81-82
Sources: README.md81-82
A pre-processing layer that analyzes the user's natural language input to determine the underlying intent (e.g., refactoring, searching, or planning) before delegating to a specific agent.
keyword-detector hook packages/omo-opencode/src/hooks/AGENTS.md1-10
Sources: packages/omo-opencode/src/hooks/keyword-detector/AGENTS.md1-10 packages/omo-opencode/src/hooks/AGENTS.md1-10The primary orchestrator agent. Its role is to break down complex tasks, delegate them to specialized subagents, and verify the results.
Specialized agents that perform specific types of technical tasks:
A subgroup of agents dedicated to the strategy phase:
The following diagram bridges natural language agent roles to their specific code identifiers and typical model targets defined in the requirements registry.
Agent Mapping: Natural Language to Code Entity
Sources: docs/reference/features.md11-31 docs/guide/agent-model-matching.md43-69 packages/omo-opencode/src/agents/AGENTS.md1-10
An asynchronous execution unit managed by the BackgroundManager. These tasks allow agents to run in the background while the user continues to interact with the primary session.
BackgroundManager handles the lifecycle, polling, and concurrency limits packages/omo-opencode/src/features/background-agent/AGENTS.md1-10startAttempt, finalizeAttempt, and bindAttemptSession packages/omo-opencode/src/features/background-agent/AGENTS.md1-10ConcurrencyManager which implements queue-per-key architecture packages/omo-opencode/src/features/background-agent/AGENTS.md1-10
Sources: packages/omo-opencode/src/features/background-agent/AGENTS.md1-10A code manipulation technique that uses LINE#ID content hashing to ensure that an agent's edit is applied to the exact intended line, preventing "hallucinated" edits on shifted line numbers.
hashline_edit config flag assets/oh-my-opencode.schema.json107-109createHashlineReadEnhancerHook augments file reading with hash metadata packages/omo-opencode/src/hooks/AGENTS.md1-10
Sources: assets/oh-my-opencode.schema.json107-109 packages/omo-opencode/src/hooks/AGENTS.md1-10An internal term for the autonomous execution loop that powers ultrawork. It ensures that if an agent stops prematurely or fails, the system automatically resumes the task.
ulw-loop and cancel-ralph commands assets/oh-my-opencode.schema.json76-77createRalphLoopHook manages the state transition of the loop packages/omo-opencode/src/hooks/AGENTS.md1-10
Sources: assets/oh-my-opencode.schema.json76-77 packages/omo-opencode/src/hooks/AGENTS.md1-10The todoContinuationEnforcer hook. It acts as a "weight" (like Sisyphus's boulder) that prevents the session from ending if there are incomplete tasks in the TODO list.
boulder-state package package.json29createTodoContinuationEnforcer packages/omo-opencode/src/hooks/AGENTS.md1-10
Sources: package.json29 packages/omo-opencode/src/hooks/AGENTS.md1-10A mapping system that associates task types (e.g., visual-engineering, ultrabrain) with specific models and prompt appends.
categories section of the configuration docs/reference/configuration.md99-124quick, unspecified-low, unspecified-high, writing, and visual-engineering docs/reference/configuration.md99-124
Sources: docs/reference/configuration.md99-124The initialization sequence for the oh-my-openagent plugin.
Plugin Initialization Flow
Sources: AGENTS.md45 packages/omo-opencode/src/features/AGENTS.md1-10 packages/omo-opencode/src/hooks/AGENTS.md1-10 packages/omo-opencode/src/tools/AGENTS.md1-10
A discovery mechanism for additional capabilities (MCPs or standalone tools) that can be embedded into agents.
skills-loader-core package package.json28disabled_skills in the configuration assets/oh-my-opencode.schema.json45-63
Sources: package.json28 assets/oh-my-opencode.schema.json45-63A multi-agent collaboration system where a lead agent coordinates several member agents in isolated worktrees.
team-core package package.json23| Abbreviation | Full Term | Description |
|---|---|---|
| OmO | Oh My OpenCode | The primary plugin identifier and legacy package name package.json2-3 Also dual-published as oh-my-openagent during the rename transition README.md132-133 |
| ulw | Ultrawork | The command for autonomous multi-agent execution README.md109 |
| MCP | Model Context Protocol | Standard for tool and server integration package.json12 |
| LSP | Language Server Protocol | Used for code intelligence tools (goto, symbols, diagnostics) package.json15 |
| AST | Abstract Syntax Tree | Used for structural code search (e.g., via ast-grep tools) package.json10-11 |
| JSONC | JSON with Comments | The standard format for plugin configuration files, supporting comments and trailing commas docs/reference/configuration.md59-60 |
| Sources: package.json2-15 README.md109 README.md132-133 docs/reference/configuration.md59-60 |
Refresh this wiki
This wiki was recently refreshed. Please wait 1 day to refresh again.