seed

Scanning repository documentation like CLAUDE.md and markdown prompts automatically extracts trigger phrases for model-scoped recall. It skips already scanned files unless a rescan is requested, merging new JSON triggers into the recall index.

Trigger discovery seeding automatically indexes prompt files and project markdown to build a master recall index. It applies batched researcher analysis to extract model-scoped triggers, eliminating manual hand-tagging across the codebase.

Yes, built-in skipped-file detection prevents reprocessing already scanned markdown and prompt files. It only reprocesses previously indexed documentation if an explicit repository rescan is requested by the user.

To bootstrap a new project, scan the repository path to generate model-scoped trigger phrases from documentation. The system merges extracted triggers into a JSON file and rebuilds the master recall index for immediate searchability.

Yes, the scanner indexes CLAUDE.md docs, agent commands, skills, and readmes for immediate recall. It applies batched researcher analysis to these markdown prompts and rebuilds the recall index for model-scoped discovery.