docs: sync 2026-08-29 - #496
Open
JulienMartel wants to merge 1 commit into
Open
Conversation
Docs-Sync: 2026-08-29
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
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Docs sweep — reconciled 9 commits here, and carries the findings for four repos whose docs needed nothing.
scruff got a tree on hausfold.co yesterday (hausfold.co #183) and cut its README down to a door (scruff #92), deleting
docs/lifecycle.mdanddocs/reference.md. Two places here still say the opposite, and one of them is the routing table the next sweep reads.Corrected
AGENTS.md▸ Where a write-up goes — was: "haus, pounce and perch each have one" tree, and "scruff, snug and nebelung have no tree at all, so their README is the front page … never cut one of those down to a door." Both halves are now wrong about scruff, and the second is an instruction to undo what scruff notes: Phase 3 mostly done — light mode, keys.*, and the readiness test passing #92 did. scruff joins the first group; the second is snug and nebelung. Also names the two things that legitimately stayed in scruff's repo, since the first group's rule is "what a contributor needs":docs/releasing.md, andSPEC.md, which sits at the root rather than indocs/..agents/skills/docs-sync/SKILL.md▸ Step 2 — thescruff/row said "scruff has no site tree, so 'the docs' for it are the repo's", which would send every future sweep to the wrong surface. It now routes to the tree and names what stays behind, plus the three-file rule scruff's ownAGENTS.mdstates (a verb, flag, exit code or--jsonkey moves the code,ai/SKILL.mdand the page together). Theorg-profilerow's "only the four trees" is five.Left alone (deliberately)
bench statusnesting a spawned lane under its parent (status: nest a spawned lane under the lane that spawned it #494) landed with its owndocs/workflows.mdedit in the same commit. Nothing to add.docs/factory.md, the/nightshiftskill and its README and AGENTS.md rows in one go. Reconciled by construction.org-profilehad no commits in range, so its README was not re-audited — but one thing is worth naming rather than fixing: it lists seven repos andsnugis not one of them. Its own AGENTS.md says the README is "one line per repo, in the family's order (house → apps → theme → substrate → bench)", and snug is a public family repo that this repo's README already carries. Not written, because nothing in the range put it there and I can't tell from the history whether the omission is a decision. Would change on a word either way.holtspellings that survive (scruff'sSPEC.md,sdkdocs,.github/copilot-instructions.md,haus/AGENTS.md's two-release migration note,/release's SKILL.md) are all provenance for immutable published artifacts — Go proxy paths and PyPI names that resolve forever at the tags published under them — or the named pattern for the next prefix move. Kept, per the "names a shape rather than narrating a file's past" carve-out.Needs code, not docs
stageorrankingsettings.modules/launcher/default.nix:1470writes~/.config/pounce/config.jsonfrom a fixed key list with neither block in it and no passthrough;pounce config initrefuses inside haus and the Settings window opens read-only. pounce docs:wt park/wt unpark— the stash-free way to set work aside #123/docs: swap layouts are Alt [ / Alt ], and now work locked #124 added fourranking.*switches, all defaulting on, all of which change what⌘Spaceshows before you type — and hausfold.co#187 now documents them for a machine that has no way to set them. Wantshaus.launcher.*options, or one escape hatch, in haus.content/docs/haus/rooms/ai.mdxholds 40 em dashes against that repo's own "no em dashes in reader-facing copy, anywhere" rule. hausfold.co#187 leaves them (its own added lines carry none) because a 40-instance sweep of one page is its own change, and the same repo's "sweep a spelling everywhere at once, or not at all" says not to do it per-page.Where the rest of the range went
READMEdoors,assets/README.md,docs/rules.md, snug'sREADME/AGENTS.mdfor the SIGINT contract), and the site absorbed the rest.opscould not be cloned (private; the container has no credentials for it), so its half of the sweep did not run. Its watermark is untouched, so the next run with credentials picks the range up whole.Generated by Claude Code