Researches open-source GitHub repositories using parallel codebase search queries.
npx skills add https://github.com/Ohmnia/site-build --skill code-research-ohmnia
Please help me install this Agent Skill. Skill: code-research Source: https://github.com/Ohmnia/site-build/tree/main/.opencode/skills/code-research Command: npx skills add https://github.com/Ohmnia/site-build --skill code-research-ohmnia
github_codebase_search queries concurrently to minimize latency. - Implementation Analysis: Locates entry points, key functions, data flow, and configuration for any feature in a public GitHub repository. - Use Case: Before integrating an authentication library, ask how vercel/next.js handles session management. The Skill runs parallel queries on middleware, API routes, and token validation, then returns a cross-referenced summary with exact file paths. ## Quick Start Ask how a specific open-source repository implements a feature, for example: research how facebook/react implements its reconciliation algorithm.High-intent search queries and answers about installing and using this skill.
Identify the GitHub repository, then break your question into 2-4 focused search angles such as entry points, data flow, and configuration. Run all github_codebase_search queries in parallel and cross-reference the returned files to build a complete answer.
Use the github_codebase_search tool with targeted queries like 'Find the authentication middleware and session handling'. Running multiple queries concurrently covers different angles of the codebase faster than sequential searches.
Parallel queries minimize latency and cover independent aspects of a feature simultaneously, such as API routes, middleware, and configuration. Sequential searching wastes time and can miss cross-cutting relationships between components.
Yes, run the same set of focused queries against each repository and compare the returned implementations side by side. This works well for evaluating architectural patterns before choosing a dependency.
Codebase search only covers public GitHub repositories and depends on query quality, so vague questions return shallow results. It also cannot execute code, so runtime behavior must be inferred from static source analysis.
LayerDynamicsAnalyze code files to formulate targeted clarifying questions about scope and integration.
gigaverse-appDiscover and reuse established code patterns across codebases.
z23ccCompile version-specific docs, source references, and known issues for frameworks.
opendatahub-ioCreates a depth-1 clone of a Git repository in a temporary location.