docs: gitignore guidance for the derived index; release 0.2.0 (#8) - #12
Merged
Conversation
Nothing in the repo told users not to commit <bundle>/.index/ -- grep for "gitignore" across every .md, .py and .json returned nothing. The only implicit guidance was this repo's own .gitignore, which ignores knowledge/: the whole bundle directory, under a hardcoded name. As a pattern to copy that is both too narrow (only a bundle literally called knowledge) and too broad (excludes the concepts, which are what you want in git). Ships templates/gitignore-fragment and a README note beside the build command, using **/.index/ rather than */.index/ so it holds at any nesting depth and for every bundle. This repo's own .gitignore now demonstrates the pattern it recommends. The README code block also stops saying knowledge/.index/ and says <bundle>/.index/, since the hardcoded name is what caused the confusion. A repo with two bundles is where the narrower pattern shows: ignoring knowledge/.index/ leaves the second bundle's index sitting in git status. Also adds the first CHANGELOG.md this repo has had, covering all seven fixes in 0.2.0, and bumps the four version manifests. There is no test enforcing version consistency here, so all four were checked by hand. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_011XXDjM9WnCtsdRTU441WUu
The release bump landed in package.json and 3 other manifests but left 8 more stale at 0.1.0, with nothing to catch it. Bump the remaining plugin manifests (opencode, codex, grok marketplace x2, catalog.json), README, user guide, and design doc, plus a UI version badge the sweep turned up. Add TestVersionConsistency to tests/test_dekc.py so a future partial bump fails CI instead of shipping silently.
Clean auto-merge except a duplicate 'import re' left by the union of independently-added imports; collapsed to one.
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.
#8 — nothing told users not to commit the index
grep -rn gitignoreacross every.md,.pyand.jsonin the repo returnedno hits: no README mention, no docs mention, no init step, and no template.
The only implicit guidance was this repo's own
.gitignore, which ignoresknowledge/— the whole bundle directory, under a hardcoded name. As a patternto copy that is simultaneously too narrow (only a bundle literally called
knowledge) and too broad (excludes the concepts themselves, which are exactlywhat you want in git).
A repo with two bundles is where it shows: ignoring
knowledge/.index/leavesthe second bundle's index sitting in
git statusas untracked, which is how Iran into it.
Changes:
templates/gitignore-fragmentships with the plugin.**/.index/rather than*/.index/, so it holds at any nesting depth and forevery bundle rather than one name.
.gitignorenow demonstrates the pattern it recommends.<bundle>/.index/instead ofknowledge/.index/—the hardcoded name is what caused the confusion in the first place.
Release 0.2.0
This repo had no
CHANGELOG.md. This adds the first one, covering all sevenfixes that make up 0.2.0, and bumps the four version manifests
(
marketplace.json,.claude-plugin/marketplace.json,.claude-plugin/plugin.json,package.json).There is no test enforcing version consistency here — unlike
okf-plugin, whichhas one — so all four were checked by hand. That might be worth adding later.
Depends on #9, #10 and #11 landing first, since the changelog describes them.
Verification
Closes #8