Query project reference documentation repos using semantic search and RAG answers.
npx skills add https://github.com/contexthub-md/marketplace --skill kb-contexthub-md
Please help me install this Agent Skill. Skill: kb Source: https://github.com/contexthub-md/marketplace/tree/main/contexthub/skills/kb Command: npx skills add https://github.com/contexthub-md/marketplace --skill kb-contexthub-md
ch_discover_context for factual lookups and browsing. - RAG Answers: Get synthesized, sourced answers to questions like "how do I create a Strands agent tool?" with citations from the docs. - Docs-to-Code Cross-Reference: Bridge documentation with your actual codebase using ch_ocxp_search and ch_ocxp_grep to see how documented patterns are used in practice. - Use Case: You are integrating EMQX authentication and ask "what MQTT QoS levels does EMQX support?" — the Skill searches the indexed EMQX docs, returns a sourced answer, then shows how your codebase already configures QoS. ## Quick Start Ask the kb skill a question like "how do I configure Strands agent memory?" to search the project's reference docs and get a sourced answer.High-intent search queries and answers about installing and using this skill.
Use ch_discover_context with your query and content_type set to docs to retrieve relevant chunks from indexed reference repos. For a synthesized answer with sources, set include_answer to true to enable RAG mode.
The kb skill targets REFERENCE-tagged repos such as library docs, framework guides, and protocol specs. Use the discover skill for general project-knowledge search not tied to reference repos, and the research skill for external library docs or web search.
Yes, RAG mode generates a sourced answer. Call ch_discover_context with include_answer set to true, and it retrieves relevant doc chunks and produces an answer citing the sources it drew from.
The kb skill requires an active connected project with indexed knowledge base repos. Run the connect skill first to set up your workspace and project before querying the knowledge base.
After finding relevant docs, use ch_ocxp_search with the topic or ch_ocxp_grep with a specific API or class name from the docs. This shows how documented patterns are actually implemented in your codebase.
MrGKanevIndex Markdown notes locally and search via keyword, semantic, or hybrid modes.
jamesgray-aiGenerate Standard Operating Procedures for manual, augmented, and automated workflows.
lucaslessachavesFetch Databricks documentation references via the llms.txt index.
mikebronnerPlan and execute multi-source searches with result curation and documentation.