Add research-orchestrator skill

What problem it solves

Research tasks in Claude Code today are single-threaded: one agent works through a question linearly, which is slow for complex, multi-faceted queries. research-orchestrator solves this by decomposing a research question into independent workstreams, dispatching parallel sub-agents, monitoring their progress, and synthesizing their findings into a single report with inline citations.

Who uses this workflow

Anyone doing deep research with Claude Code -- competitive analysis, technical due diligence, literature reviews, market landscaping -- where the question naturally breaks into multiple independent lines of inquiry that can run simultaneously.

What the skill does

• Decomposes complex research questions into parallel, independent workstreams
• Dispatches and monitors sub-agent progress across workstreams
• Synthesizes results into a unified report with inline citations
• Handles failures gracefully (kills stuck agents, relaunches with pre-loaded data)

Category

Data & Analysis -- alongside existing research-oriented skills like deep-research and root-cause-tracing.

Links

• Repository: https://github.com/altmbr/claude-research-skill