docs(agents): document Oh My OpenCode multi-agent framework (closes #152)

Author: d-morrison

chapters/ai-tools/oh-my-opencode.qmd

@@ -0,0 +1,45 @@
+[`opensoft/oh-my-opencode`](https://github.com/opensoft/oh-my-opencode)
+(often referenced as **Oh My OpenCode** or `omo`)
+is an open-source multi-agent orchestration framework and plugin
+for the OpenCode coding agent harness
+(measured 2026-09-01).
+Inspired by modular terminal configuration frameworks (such as Oh My Zsh),
+it expands single-agent coding into a specialized multi-agent system
+with automated model routing and background task execution.
+
+#### Hub-and-spoke agent architecture
+
+Rather than relying on a single generalist agent to perform all research, design, coding, and review steps,
+Oh My OpenCode implements a hub-and-spoke delegation model
+with specialized agent personas:
+
+- **Sisyphus (Primary Orchestrator)**:
+  Acts as the lead coordinator that breaks user requests into discrete work packages,
+  delegates subtasks to specialized agents, and aggregates results into a coherent solution.
+- **Oracle (Strategic Advisor & Architect)**:
+  Focuses on architectural planning, system design tradeoffs,
+  and reviewing complex implementation strategies before code changes are made.
+- **Librarian (Research & Codebase Navigator)**:
+  Conducts fast codebase exploration, documentation lookups,
+  and dependency inspection to supply targeted context without bloating the orchestrator's context window.
+- **Designer (UI/UX Specialist)**:
+  Specializes in frontend component architecture, styling conventions,
+  and interactive UI workflows.
+
+#### Heterogeneous model routing
+
+A central capability of Oh My OpenCode is decoupling agent roles from a single model provider.
+Developers can route distinct tasks to the model family best suited for each workload:
+
+- **High-reasoning tiers** (such as Claude Sonnet or OpenAI reasoning models) for architectural planning and refactoring.
+- **High-throughput or large-context models** (such as Google Gemini) for extensive repository research and log analysis.
+- **Lightweight local or specialized models** (via Ollama or OpenRouter) for routine linting and documentation lookups.
+
+#### Operational capabilities
+
+| Dimension | Standard OpenCode | Oh My OpenCode |
+|---|---|---|
+| **Agent topology** | Single sequential agent loop | Hub-and-spoke multi-agent team |
+| **Model routing** | Single active model per session | Dynamic per-agent model assignment |
+| **Execution monitoring** | Standard terminal output | Background execution and tmux session multiplexing |
+| **Extensibility** | Individual plugins and MCPs | Curated bundle of tools, agents, and MCP integrations |

chapters/coding-agents.qmd

@@ -185,6 +185,10 @@ to [help you code](https://en.wikipedia.org/wiki/AI-assisted_software_developmen
 
 {{< include ai-tools/magic-context.qmd >}}
 
+# Multi-Agent Orchestration with Oh My OpenCode {#sec-ai-oh-my-opencode}
+
+{{< include ai-tools/oh-my-opencode.qmd >}}
+
 # Managing Gemini API Spend and Cost Optimization {#sec-ai-gemini-spend-management}
 
 {{< include ai-tools/gemini-spend-management.qmd >}}

inst/WORDLIST

@@ -84,6 +84,7 @@ RevealJS
 SERG
 SWE
 SemBr
+Sisyphus
 Stanford
 SuperGrok
 TORQCLAW
@@ -128,6 +129,7 @@ mistyped
 nano
 navbar
 omics
+omo
 opencode
 orchestrator
 orchestrators
@@ -151,6 +153,7 @@ submodule
 symlink
 symlinked
 symlinks
+tmux
 toolset
 toolsets
 tidyverse