Search Brain knowledge artifacts for captured decisions, people, plans, and sources.
npx skills add https://github.com/reddb-io/red-skills --skill search-reddb-io
Please help me install this Agent Skill. Skill: search Source: https://github.com/reddb-io/red-skills/tree/main/plugins/brain/skills/core/search Command: npx skills add https://github.com/reddb-io/red-skills --skill search-reddb-io
brain_search MCP tool or the red-skills-brain search CLI, returning hits ranked by lexical matches, tags, artifact kind, and graph connections. - Transparent Relevance Scoring: Exposes a score_breakdown per hit so the agent can explain why a result ranks high and cite artifact id, kind, and title. - Brain-vs-Memory Routing: Directs operational engineering facts (gotchas, why-notes) to /memory:recall while keeping human and project knowledge in Brain. - Use Case: A user asks "what do we know about the Acme migration decision?" The agent searches Brain, cites the matching decision artifact with its id and score breakdown, and notes any gaps instead of inventing details. ## Quick Start Ask the agent to search the brain for what was captured about a specific decision, person, or open question before answering.High-intent search queries and answers about installing and using this skill.
Call the brain_search MCP tool with a natural-language query and a result limit, or run red-skills-brain search from the CLI. Results are ranked by lexical matches, tags, artifact kind, and graph connections, and each hit includes a score_breakdown.
Brain search retrieves human and project knowledge such as decisions, people, ideas, and plans. Memory recall (/memory:recall) retrieves operational engineering facts like gotchas, why-notes, and validated approaches from work sessions. When unsure, search Brain first.
Use brain search when you want the raw ranked hit list of matching artifacts. Use /brain:think when you want a synthesized, cited answer drawn from multiple hits, since brain_think runs search internally and returns a grounded answer with confidence and evidence gaps.
State plainly that Brain has no captured knowledge about the query and do not substitute model knowledge as if it came from Brain. Suggest /brain:capture so the missing knowledge can be stored for future searches.
Treat each hit as a claim made at capture time and verify it still holds before relying on it for high-stakes actions. Cite the artifact id, kind, and title so the user can reference and check the source.
garrytanSearches a personal knowledge brain for references and stored information.
EvolutionAPIAggregate internal and external sources to answer customer questions with attributed references.
brolagStore and recall user-provided facts across sessions with tiered memory architecture.
Kingly-AgencyBuild RAG pipelines that ground LLM responses with retrieved external knowledge.