Tags: akhayam99/goodboy
Tags
chore(repo): bump version to 0.1.82 (#1423) * chore(repo): bump version to 0.1.82 Bump the six version carriers to 0.1.82 and add the v0.1.82 changelog section, which the release build reads as the release body. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> * docs(repo): keep the v0.1.82 notes to observable behaviour Drop the ordering detail from the Slack connect paragraph and state the failed-connect outcome instead, and drop the render-timing wording from the checklist paragraph. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> * docs(repo): repair the v0.1.82 notes after review Voice steward rewrites: the fifth checklist item is named an agent to match its shipped title, the two screens sharing a checklist entry are named as the tools step and the Sentry step rather than by a source filename, and the renamed step reads as now called rather than as a title ending in now. Challenger corrections: the lead line leads with the fast-forward and the first-run theme, the git section heading names the capability in on-screen words, the failed reads named are the three the app classifies as failures, the in-progress operations line drops an invented before-state, the Slack rollback claims the restored record and not a working integration, the confirm targeting line is cut because no released build had that hazard, and the plain-words fix line is cut because the panel still shows git vocabulary in two labels. Adds a line for the shipped documentation corrections, a declared departure from the runbook filter, and the follow-up line the tone doc requires for the fast-forward. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> --------- Co-authored-by: Dev <demo@example.com> Co-authored-by: Claude Opus 5 <noreply@anthropic.com>
chore(repo): bump version to 0.1.81 (#1409) * chore(repo): bump version to 0.1.81 Six version places plus the CHANGELOG section the release build reads. Notes drafted from the merged PR bodies of #1405, #1406 and #1407, with every claim re-derived against the merged diffs at 31b6321. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> * docs(repo): correct six release-note claims the challenger falsified Phase 8 challenger pass over the draft notes, each correction re-derived against origin/main at 31b6321 rather than taken from the report: - drop the max mode keys bullet: #1405's body says boot seeded authResults, so clearAll never fired in any shipped build. It advertised a fix for a bug no user had. - one to seven and a half seconds, not one to seven: the baseline table reads 1084-7514 ms, and the headline understated its own evidence. - the accent bullet named one colour for two assets: the desktop token is oklch(... 220) and the website's is oklch(... 200), and the favicon was rendered from the site's. - repository creation runs through gh repo create and gh repo view, not a published REST API. - heading rewritten: failures do not name anything, the screen does. - cut the focus-refresh clause as mechanism, and state the boot error screen's unshortened report link inside the paragraph that promises the crash screen shortens home folders. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> * docs(repo): take the voice steward's blocking rewrite on the notes Phase 8 voice steward pass over the draft notes and the announcement post. Blocking, taken verbatim: "fourteen provider checks" is the internal shape of the old boot path, the same category tone-of-voice.md excises in its own fix-line example. The timing stays, because that is what the reader felt. Improvement, taken with one change: the boot error screen's unshortened report link now sits inside the sentence that promises the crash screen shortens home folders, rather than as its own clause at the end. The steward's proposed wording used an em dash, which the same law bans, so it reads "though" instead. Improvement, declined with a reason: splitting that paragraph after "you submit the form there" would take the section to three content paragraphs, and the steward passed it at two against the length budget. The harder rule wins. The post takes "seven and a half seconds" for consistency with the notes, 271 characters against a 280 budget. It is not published: the body and the reply are reviewed, the image gate and publication are the lead's. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> --------- Co-authored-by: Dev <demo@example.com> Co-authored-by: Claude Opus 5 <noreply@anthropic.com>
chore(repo): bump version to 0.1.79 (#1393) * chore(repo): bump version to 0.1.79 * docs(repo): correct four release note claims against the diff The Phase 8 challenger read the notes cold against the diffs and found four defects, all fixed here. The impact panel note said sessions that finished before this release stay absent. False: sessionGithub is never hydrated at boot, so the first sweep after updating cannot take the merged-or-closed skip and refreshes every branched session, writing a row for each. The consolidation bullet said behavior is unchanged, on a 271 file diff that reroutes four hand-rolled dropdown backdrops, drops two re-implemented popovers and moves layering values onto tokens. Written from the PR body rather than the diff. The resolver note said hint actions went missing on fix and rework and are back. They were rendered before the change. The disappearance was real but on the busy and locked axis, so the note named the wrong one. The spend limit note stated the default and buried the consequence: the fixed step ceiling is deleted, so a dynamic run has no automatic stop until a user sets one. Also: two headings renamed off speech verbs, two features trimmed to the length budget, and the lead line's third clause replaced.
docs(repo): correct one release-note claim before the tag (#1368) The challenger's notes review found the #1361 line half true: secret_get returned the credential, secret_has only reported whether one existed. Corrected before the tag, because release.yml reads this section verbatim into both the GitHub release body and the in-app updater. The follow-up line is also rewritten on the same review: it carried process jargon and closed on a circular sentence describing what a test does rather than what the reader should do.
PreviousNext