What problem does it solve? Single-model research often misses perspectives and produces biased or incomplete answers. This Skill runs the Discover phase of the Double Diamond workflow by dispatching a research question to multiple external AI providers (Codex, Antigravity, Copilot, Qwen, Ollama, Perplexity, OpenRouter, and others) in parallel, then synthesizing their outputs into one structured, source-attributed report. ## Core Features & Use Cases - Dynamic provider fleet: Uses build-fleet.sh to detect locally available providers and assign diverse perspectives, enforcing model-family diversity to reduce agreement bias. - Adjustable research intensity: Supports quick (2 agents), standard (4-5 agents), and deep (6-10 agents) modes via breadth/intensity parameters. - Structured synthesis: Produces a report with executive summary, key findings, themes, conflicts, gaps, priority matrix, recommendations, and per-provider source attribution, saved as a probe-synthesis markdown file. - Use Case: Ask "Compare Redis vs Memcached for session storage" and receive a multi-provider comparison covering implementation details, ecosystem trends, and trade-offs, with each claim attributed to its source provider. ## Quick Start Ask the assistant to research a topic with the octo discover workflow, for example: run octo research on OAuth 2.0 authentication patterns for React apps.
