What problem does it solve?
Repeatedly rereading raw source files and documentation into conversation context wastes tokens and slows down AI-assisted work. This Skill queries a persistent FTS5 index of previously indexed local projects, documentation, and session memory so answers come from the index instead of reloading raw sources.
Core Features & Use Cases
- Indexed Full-Text Search: Query the context-mode FTS5 knowledge base via the
ctx_search MCP tool or the context-mode search CLI fallback.
- Scoped and Batched Queries: Scope searches with
source labels like project:<name> and batch multiple related questions in a single queries array.
- Use Case: While working on a large codebase, ask how authentication middleware and token refresh are implemented; the Skill returns relevant indexed snippets without rereading the entire source tree into context.
Quick Start
Ask the assistant to search the context-mode index for how authentication middleware works in your project.