fix(claude-memory): re-quote settings-page blockquotes verbatim in stateless reference - #2040
Conversation
…ateless reference The precedence quote in reference/official-guidance.md substituted a bare ellipsis for item 1's parenthetical, putting non-page text inside an attributed blockquote. Restores the real words and records the two normalizations (links flattened to labels, nested detail bullets omitted) in the attribution note. Two further fidelity defects surfaced in the same pass. The "Cannot be overridden by any other level" quote truncated mid-sentence, dropping ", apart from the exceptions in the bullets below" and turning a qualified claim absolute; the full sentence is restored and the operative conclusion moved to prose that names the exception keys. The claude project purge quote asserted a "requires Claude Code v2.1.124 or later" sentence the page no longer carries; it is out of the quote, and the retained version floor is labelled as having no current upstream source. The env and cleanupPeriodDays quotes were re-checked character-for-character against the live page and are verbatim as they stand, so their wording is unchanged. New prose resolves cleanupPeriodDays' "session files and other application data" against the cleanup table it links to, so it no longer reads as contradicting the adjacent statement that sessions/ is not age-swept. Refs #1989 (row 247) Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
|
Bugbot is not enabled for your account, so this pull request was not reviewed. Enable Bugbot in the Cursor dashboard to get automatic reviews on future PRs. |
|
Claude finished @kyle-sexton's task in 35s —— View job Security review complete
No security issues found. All three changed files are non-executable content:
None of this touches:
This is a documentation-accuracy fix with no code paths, no attack surface, and nothing for a security review to flag. |
…d negative The prose replacing the truncated "Cannot be overridden" quote enumerated four security-sensitive keys as though that were the whole exception set. Item 1 also carries the within-managed-tier source ranking with its own exception keys, the SDK managedSettings / parentSettingsBehavior embedder path, allowManagedPermissionRulesOnly, and forceLoginOrgUUID / allowedMcpServers, so the summary under-enumerated the page — the same defect class this branch exists to fix, in the opposite direction. Replaces it with the claim the skill actually needs and can hold accurate: no exception bullet names autoMemoryEnabled, CLAUDE_CODE_DISABLE_AUTO_MEMORY, or auto memory, so no ordinary lower scope overrides a managed auto-memory value. Readers go to the page for the exception set itself. Refs #1989 (row 247) Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
![chatgpt-codex-connector[bot]](https://avatars.githubusercontent.com/in/1144995?s=60&v=4){kind=small}
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 96e1dc2b5b
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
|
Claude finished @kyle-sexton's task in 45s —— View job Security review complete
No security issues found. I reviewed
None of the changed content touches:
This is a documentation-accuracy fix (re-quoting blockquotes against live upstream docs) with no attack surface. Confirms the automated review already posted on this PR — nothing further to flag. |
|
Claude finished @kyle-sexton's task in 6m 0s —— View job Review complete
This is a code-correctness review (security is the dedicated security lane's job, which already ran and found nothing on this diff). Two inline comments posted on What I independently checked (not just the PR's own Test plan claims):
Maintainability observation (not blocking): the CHANGELOG entry notes the Not independently verified by me (author-claimed only): the Nothing else in the diff (the version bump, the CHANGELOG prose, or the rest of the re-quoted/restored text) raised correctness or architecture concerns. |
![claude[bot]](https://avatars.githubusercontent.com/in/1236702?s=60&v=4){kind=small}
|
Claude has reviewed this PR 4 times. The lane skips further automatic reviews after 5; deleting this comment resets the count. |
Review follow-ups on the official-guidance.md precedence rewrite: - The "no ordinary lower scope overrides a managed auto-memory value" conclusion overstated: it now governs settings scopes only, with the CLAUDE_CODE_DISABLE_AUTO_MEMORY environment variable called out as sitting outside settings precedence and still overriding the effective value, managed or not (per the file's own env-var section). - "longer and more varied than this file summarizes" implied a summary the file no longer gives; reworded to "than is useful to enumerate here". - The rewritten precedence quote was re-verified against the live settings page (raw markdown, 2026-08-08): "Settings apply in order of precedence" list, items 2-5, and the "Cannot be overridden by any other level ... apart from the exceptions in the bullets below" nested bullet all match verbatim; item 1's three links flatten to the labels the attribution note records; the exception bullets name auto memory nowhere. - CHANGELOG 0.5.8 entry extended with the narrowed conclusion (still unreleased; no version bump). Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01HbtPzLRe1yBavNpsmv5Tum
|
Claude finished @kyle-sexton's task in 1m 17s —— View job Security review complete
No security issues found. I diffed this PR against its actual base commit (
None of this diff touches:
This is a documentation-accuracy fix with no attack surface. This confirms the prior automated security-lane reviews already posted on this PR — nothing further to flag from a security standpoint. (Note: this run did not check |
|
Claude finished @kyle-sexton's task in 4m 17s —— View job Code review complete
Scope: correctness/maintainability review of the diff ( One inline finding posted on The PR's stated intent is to label the On the still-open precedence-quote verification (flagged unresolved by the prior claude review round, comment): I independently re-attempted this via Everything else in the diff: the codex P2 finding (managed-settings negative needing to exclude the OS env var) and the prior claude wording nit ("longer and more varied than this file summarizes") were both fixed in |
Review follow-ups on official-guidance.md: - The storage-location section restated the v2.1.124+ purge floor as a bare doc-backed fact 75 lines before the Out of scope section labels the same claim "no current upstream source". The floor is now stated once, at the labelled site; the earlier mention points there, flagging that the sourcing status travels with it. - "neither does cli-reference" asserted a negative about a page absent from the file's Sources list, with no URL or verification stamp. Verified against the live page (raw markdown, 2026-08-08): cli-reference documents `claude project purge` with no version requirement anywhere. The negative now names the page with a verification date, and cli-reference joins the Sources list and the header's verified-2026-08-08 set. - CHANGELOG 0.5.8 entry extended (still unreleased; no version bump). Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01HbtPzLRe1yBavNpsmv5Tum
|
Claude finished @kyle-sexton's task in 35s —— View job Security review complete
No security issues found. I independently pulled the full diff via
None of this diff touches, or introduces:
This is a documentation-accuracy fix with no attack surface. Confirms all prior security-lane reviews on this PR — nothing further to flag from a security standpoint. |
|
Claude finished @kyle-sexton's task in 2m 22s —— View job Code review complete
Scope: correctness/maintainability only (security already cleared by the dedicated security lane, 3 times over). This is review round 5 — I focused on whether the newest commit actually closed out what round 4 flagged, plus an independent pass over the rest of the diff, rather than re-litigating settled ground. Round-4 finding is resolved, correctly, by
Independently confirmed (read myself, not author-claimed):
Not independently verified by me: the settings-page precedence-quote fidelity itself — I don't have working No new findings. Nothing in the |
Main absorbed PR #2018, which released claude-memory 0.5.8 with different content. This PR's entry moves from 0.5.8 to 0.5.9, above main's 0.5.8 entry; plugin.json follows. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01HbtPzLRe1yBavNpsmv5Tum
|
Claude finished @kyle-sexton's task in 1m 25s —— View job Code review complete
Scope: correctness/maintainability only (security is the dedicated security lane's job and has already cleared this diff four times). This is review round 6, triggered by a new commit since the last review ( What's new since the last review round: Independently confirmed (read myself):
Not independently verified by me: No new findings. The merge/renumber is mechanically sound and doesn't touch the content prior rounds already scrutinized. |
|
Claude finished @kyle-sexton's task in 47s —— View job Security review complete
No security issues found. This trigger fired on the latest push (
None of this diff touches, or introduces:
This is a documentation-accuracy fix with no attack surface. Confirms all prior security-lane reviews on this PR — nothing further to flag from a security standpoint. |
1 participant
No linked issue
This fixes one row of the audit issue cited under
## Related; that issue stays open for itsremaining rows, so no closing keyword.
Summary
plugins/claude-memory/skills/stateless/reference/official-guidance.mdcarried attributedblockquotes whose text is not on the cited pages. This is a re-quoting pass against the live
code.claude.com/docs/en/settings.mdandcode.claude.com/docs/en/claude-directory.md, fetched2026-08-08.
Fixed — the flagged precedence quote. Item 1 read
1. **Managed settings** (…). A bareellipsis stood inside quote marks where real page words belong. It now reads
1. **Managed settings** (server-managed, MDM/OS-level policies, or managed settings), and theattribution note records the two normalizations: the three parenthetical links are flattened to
their labels (the file's existing convention for every other quote), and each item's nested detail
bullets are omitted.
Fixed — an unflagged truncation that inverted a claim. The adjacent quote read
"Cannot be overridden by any other level, including command line arguments". The live sentence continues, apart from the exceptions in the bullets below. Truncating there turned a qualified statementinto an absolute one. The full sentence is restored. The conclusion the skill actually needs is
carried by prose stating a verified negative — none of item 1's exception bullets names
autoMemoryEnabled,CLAUDE_CODE_DISABLE_AUTO_MEMORY, or auto memory at all, so no ordinary lowerscope overrides a managed auto-memory value. Deliberately not an enumeration of the exception set:
that set is long and varied (a within-managed-tier source ranking, an SDK-embedder merge path,
several admin-source key exceptions), and summarizing it here would be the same defect class in the
opposite direction. Readers go to the page for it.
Fixed — a sentence upstream has deleted. The
claude project purgequote asserted"The command requires Claude Code v2.1.124 or later."claude-directory no longer carries that sentenceand cli-reference never did. It is out of the quote, and the note beside it labels the retained
v2.1.124+floor a claim with no current upstream source. That label is added once, in this file;SKILL.mdandcontext/purge.mdstill state the floor unqualified and are deliberately untouched— sweeping a version claim across the plugin is not this row's scope, and the docs dropping the
sentence does not disprove the floor.
Unchanged — two of the three flagged quotes were already verbatim. The
envsentence and thecleanupPeriodDaysdescription were compared character-for-character against the live page andmatch exactly (modulo the same link flattening). Their wording is untouched. Manufacturing edits
to satisfy a stale premise would have made the file less accurate, not more.
Soft aggravator resolved.
cleanupPeriodDays' verbatim"session files and other application data"sat beneath prose statingsessions/is not age-swept, close enough to read ascontradicting it. The live page still carries that tension, so the quote stays verbatim and new
prose resolves the phrase against the table it links to — transcripts,
shell-snapshots/,debug/,tasks/,file-history/— notingsessions/is not a row in it, which is what thequote two paragraphs down already established.
Every settings and claude-directory verification stamp in the file moves to 2026-08-08, the date
each quote was re-checked. Version bumped to
0.5.9with a matching CHANGELOG entry. (Originally0.5.8; main then absorbed #2018, which released a different0.5.8for this plugin, so thisbranch merged
origin/mainand moved its entry to0.5.9above main's0.5.8.)Test plan
scripts/check-changelog-parity.sh --check— pass ("Every versioned plugin has a CHANGELOG.md").scripts/check-changelog-parity.sh --check-bump origin/main— pass (## [0.5.9]entry presentat head, absent at ref).
markdownlint-cli2on both changed markdown files — 0 errors.claude-directory was substring-matched against that page's fetched raw markdown with link
syntax flattened. All such quotes now match exactly; the two removals above are the strings that
did not. The file's six
memoryblockquotes and oneenv-varsblockquote were not re-checked— those pages were not fetched, re-researching them is outside this row, and the file header
already scopes its verification stamps accordingly.
Related
code.claude.com/docs/en/settingsthat are not verbatim on that page.live page as of 2026-08-08. Two additional non-verbatim quotes in the same file were found and
fixed instead.
🤖 Generated with Claude Code
https://claude.ai/code/session_01HbtPzLRe1yBavNpsmv5Tum