feat(search): hits carry file context, windowed reads, read_many - #1
Merged
Conversation
Three ergonomics fixes for the case memory is actually heading toward — more files, longer notes: - search hits include the containing file's description (H1 + first prose line, derived not authored), so a line lifted from a 30KB note is attributable without a second call - read_memory takes from_line/max_lines and reports total_lines + partial, so long notes can be read in windows - read_many reads a glob in one call with a per-file cap, replacing N round trips for 'show me all the notes' Adapted from ideas in tobi/qmd without the dependency: it loads its own in-process GGUF models, which would put a second model runtime and a second embedding universe next to the fleet's existing one. No index is added here — the substrate stays markdown in git and search stays grep. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Ideas adapted from tobi/qmd; no dependency (it runs its own in-process GGUF models — a second model runtime and second embedding universe beside the fleet's).
search_memoryhits gainabout: the containing file's H1 + first prose line, derived from content so there's nothing to author or keep in syncread_memorygainsfrom_line/max_lines, returnstotal_linesandpartialread_manytool: glob across an expert's files in one call, per-file line cap, each with its descriptionStill no index — markdown in git, search is grep.
docs/design.mdrecords the trigger for revisiting that.Tests: describe() table (title+prose / title only / prose only / heading-and-quote skipping / empty), hits-carry-description, windowed reads incl. past-end, read_many incl. truncation marking and error cases.
🤖 Generated with Claude Code