diff --git a/.changeset/always-ask-user-question.md b/.changeset/always-ask-user-question.md deleted file mode 100644 index c6d2ba2..0000000 --- a/.changeset/always-ask-user-question.md +++ /dev/null @@ -1,13 +0,0 @@ ---- -'@stijnvanhulle/template-claude-plugin': minor -'@stijnvanhulle/template-cursor-plugin': minor ---- - -The `user-questions` rule now says to always reach for the native picker (`AskUserQuestion` or -a client's equivalent) when a task is blocked on information only the user can supply, instead -of reserving it for a fixed-choice fork. - -- Covers open-ended requests too: offer your best-guess options and let `Other` carry the real - answer, rather than falling back to a plain question. -- The one carve-out is a secret or credential (an API key, a token, a password), which stays a - plain question and is never echoed back or offered as a guessable option. diff --git a/.changeset/code-reviewer-runs-slop-checks.md b/.changeset/code-reviewer-runs-slop-checks.md deleted file mode 100644 index e0dcd1d..0000000 --- a/.changeset/code-reviewer-runs-slop-checks.md +++ /dev/null @@ -1,14 +0,0 @@ ---- -'@stijnvanhulle/template-claude-plugin': minor -'@stijnvanhulle/template-cursor-plugin': minor ---- - -`code-reviewer` now folds `deslop`'s reuse-first ladder and AI style tells, `jsdoc`'s -conventions, `humanizer`'s pattern list, and `documentation`'s style guide into its review, in -that order, then asks which findings to hand off for fixing. - -- Stays read-only: it reports findings by these skills' criteria instead of applying their - fixes, then follows the `user-questions` rule to ask which of `deslop`, `jsdoc`, `humanizer`, - or `documentation` you want run next. -- A pass with nothing in its category (no exported symbols touched, no prose changed) is - skipped rather than forced. diff --git a/.changeset/deslop-runs-humanizer.md b/.changeset/deslop-runs-humanizer.md deleted file mode 100644 index 6da1b90..0000000 --- a/.changeset/deslop-runs-humanizer.md +++ /dev/null @@ -1,12 +0,0 @@ ---- -'@stijnvanhulle/template-claude-plugin': minor -'@stijnvanhulle/template-cursor-plugin': minor ---- - -`deslop` now runs the `humanizer` pattern list as a third pass over any changed README, doc, -comment block, or other user-facing markdown in the diff, alongside its existing reuse-first -ladder and code style-tell checks. - -- One `/deslop` run now covers over-engineering, code style tells, and AI writing tells in - prose, instead of needing a separate `/humanizer` pass for prose changes. -- `humanizer` still works standalone for prose review outside a code diff. diff --git a/.changeset/merge-deslop-ponytail.md b/.changeset/merge-deslop-ponytail.md deleted file mode 100644 index d7d8093..0000000 --- a/.changeset/merge-deslop-ponytail.md +++ /dev/null @@ -1,18 +0,0 @@ ---- -'@stijnvanhulle/template-claude-plugin': major -'@stijnvanhulle/template-cursor-plugin': major ---- - -Fold the `ponytail` skill and its `/ponytail` command into `deslop`, so one skill and one -`/deslop` command run both checks in a single pass. - -- Runs the reuse-first decision ladder (over-engineering: unneeded deps, wrappers, config) - alongside the existing AI style-tell checklist (needless comments, defensive checks, `any` - casts). -- Lists every finding with the file, what it violates, and the proposed fix, then asks before - changing anything, using a native picker where the client has one and a lettered list - otherwise. `deslop` used to apply its fixes directly; it now asks first, matching how - `ponytail` already worked. -- Applies only the findings you confirm. - -To upgrade, run `/deslop` where you used to run `/ponytail`. diff --git a/.changeset/reduce-skill-token-cost.md b/.changeset/reduce-skill-token-cost.md deleted file mode 100644 index 0d9e16b..0000000 --- a/.changeset/reduce-skill-token-cost.md +++ /dev/null @@ -1,13 +0,0 @@ ---- -'@stijnvanhulle/template-claude-plugin': patch -'@stijnvanhulle/template-cursor-plugin': patch ---- - -Cut the token cost of the `jsdoc` and `pr` skills without dropping any guidance. - -- Moved the `jsdoc` skill's full `@example` gallery and documentation patterns into - `references/examples.md`, so the default load only carries the tag tables, guidelines, and tag - order. -- Shortened the `jsdoc` skill's frontmatter description, since it stays in context in every - `` listing whether or not the skill loads. -- Tightened repetitive prose across the `pr` skill's steps, keeping every check and guardrail. diff --git a/packages/core/package.json b/packages/core/package.json index 5a71c41..09278f0 100644 --- a/packages/core/package.json +++ b/packages/core/package.json @@ -1,6 +1,6 @@ { "name": "@stijnvanhulle/template-core", - "version": "0.6.0", + "version": "1.0.0", "description": "Core package", "license": "MIT", "author": "Stijn Van Hulle ", diff --git a/packages/demo/package.json b/packages/demo/package.json index 2916ef1..52fd8c1 100644 --- a/packages/demo/package.json +++ b/packages/demo/package.json @@ -1,6 +1,6 @@ { "name": "@stijnvanhulle/template-demo", - "version": "0.6.0", + "version": "1.0.0", "description": "Demo package", "license": "MIT", "author": "Stijn Van Hulle ", diff --git a/tools/claude/.claude-plugin/plugin.json b/tools/claude/.claude-plugin/plugin.json index 7d1f324..099e43e 100644 --- a/tools/claude/.claude-plugin/plugin.json +++ b/tools/claude/.claude-plugin/plugin.json @@ -2,7 +2,7 @@ "$schema": "https://json.schemastore.org/claude-code-plugin-manifest.json", "name": "toolkit", "displayName": "stijnvanhulle template", - "version": "0.6.0", + "version": "1.0.0", "description": "Writing-voice skills and shared TypeScript-monorepo conventions.", "author": { "name": "stijnvanhulle", diff --git a/tools/claude/CHANGELOG.md b/tools/claude/CHANGELOG.md index 13b1fb7..1cd10eb 100644 --- a/tools/claude/CHANGELOG.md +++ b/tools/claude/CHANGELOG.md @@ -1,5 +1,63 @@ # @stijnvanhulle/template-claude-plugin +## 1.0.0 + +### Major Changes + +- [#226](https://github.com/stijnvanhulle/template/pull/226) [`e0639c2`](https://github.com/stijnvanhulle/template/commit/e0639c2936bad332a9fd7d95aa80bca796a85bab) Thanks [@stijnvanhulle](https://github.com/stijnvanhulle)! - Fold the `ponytail` skill and its `/ponytail` command into `deslop`, so one skill and one + `/deslop` command run both checks in a single pass. + + - Runs the reuse-first decision ladder (over-engineering: unneeded deps, wrappers, config) + alongside the existing AI style-tell checklist (needless comments, defensive checks, `any` + casts). + - Lists every finding with the file, what it violates, and the proposed fix, then asks before + changing anything, using a native picker where the client has one and a lettered list + otherwise. `deslop` used to apply its fixes directly; it now asks first, matching how + `ponytail` already worked. + - Applies only the findings you confirm. + + To upgrade, run `/deslop` where you used to run `/ponytail`. + +### Minor Changes + +- [#226](https://github.com/stijnvanhulle/template/pull/226) [`e0639c2`](https://github.com/stijnvanhulle/template/commit/e0639c2936bad332a9fd7d95aa80bca796a85bab) Thanks [@stijnvanhulle](https://github.com/stijnvanhulle)! - The `user-questions` rule now says to always reach for the native picker (`AskUserQuestion` or + a client's equivalent) when a task is blocked on information only the user can supply, instead + of reserving it for a fixed-choice fork. + + - Covers open-ended requests too: offer your best-guess options and let `Other` carry the real + answer, rather than falling back to a plain question. + - The one carve-out is a secret or credential (an API key, a token, a password), which stays a + plain question and is never echoed back or offered as a guessable option. + +- [#226](https://github.com/stijnvanhulle/template/pull/226) [`e0639c2`](https://github.com/stijnvanhulle/template/commit/e0639c2936bad332a9fd7d95aa80bca796a85bab) Thanks [@stijnvanhulle](https://github.com/stijnvanhulle)! - `code-reviewer` now folds `deslop`'s reuse-first ladder and AI style tells, `jsdoc`'s + conventions, `humanizer`'s pattern list, and `documentation`'s style guide into its review, in + that order, then asks which findings to hand off for fixing. + + - Stays read-only: it reports findings by these skills' criteria instead of applying their + fixes, then follows the `user-questions` rule to ask which of `deslop`, `jsdoc`, `humanizer`, + or `documentation` you want run next. + - A pass with nothing in its category (no exported symbols touched, no prose changed) is + skipped rather than forced. + +- [#226](https://github.com/stijnvanhulle/template/pull/226) [`e0639c2`](https://github.com/stijnvanhulle/template/commit/e0639c2936bad332a9fd7d95aa80bca796a85bab) Thanks [@stijnvanhulle](https://github.com/stijnvanhulle)! - `deslop` now runs the `humanizer` pattern list as a third pass over any changed README, doc, + comment block, or other user-facing markdown in the diff, alongside its existing reuse-first + ladder and code style-tell checks. + + - One `/deslop` run now covers over-engineering, code style tells, and AI writing tells in + prose, instead of needing a separate `/humanizer` pass for prose changes. + - `humanizer` still works standalone for prose review outside a code diff. + +### Patch Changes + +- [#226](https://github.com/stijnvanhulle/template/pull/226) [`e0639c2`](https://github.com/stijnvanhulle/template/commit/e0639c2936bad332a9fd7d95aa80bca796a85bab) Thanks [@stijnvanhulle](https://github.com/stijnvanhulle)! - Cut the token cost of the `jsdoc` and `pr` skills without dropping any guidance. + + - Moved the `jsdoc` skill's full `@example` gallery and documentation patterns into + `references/examples.md`, so the default load only carries the tag tables, guidelines, and tag + order. + - Shortened the `jsdoc` skill's frontmatter description, since it stays in context in every + `` listing whether or not the skill loads. + - Tightened repetitive prose across the `pr` skill's steps, keeping every check and guardrail. + ## 0.6.0 ### Minor Changes diff --git a/tools/claude/package.json b/tools/claude/package.json index 299c81c..50d5bdb 100644 --- a/tools/claude/package.json +++ b/tools/claude/package.json @@ -1,6 +1,6 @@ { "name": "@stijnvanhulle/template-claude-plugin", - "version": "0.6.0", + "version": "1.0.0", "private": true, "description": "Claude Code plugin for this template (agents, commands, skills, output styles).", "license": "MIT" diff --git a/tools/cursor/.cursor-plugin/plugin.json b/tools/cursor/.cursor-plugin/plugin.json index 397d301..83bc7ee 100644 --- a/tools/cursor/.cursor-plugin/plugin.json +++ b/tools/cursor/.cursor-plugin/plugin.json @@ -1,7 +1,7 @@ { "name": "toolkit", "displayName": "stijnvanhulle template", - "version": "0.6.0", + "version": "1.0.0", "description": "Writing-voice skills and shared TypeScript-monorepo conventions.", "author": { "name": "stijnvanhulle", diff --git a/tools/cursor/CHANGELOG.md b/tools/cursor/CHANGELOG.md index 7650a54..ab27492 100644 --- a/tools/cursor/CHANGELOG.md +++ b/tools/cursor/CHANGELOG.md @@ -1,5 +1,63 @@ # @stijnvanhulle/template-cursor-plugin +## 1.0.0 + +### Major Changes + +- [#226](https://github.com/stijnvanhulle/template/pull/226) [`e0639c2`](https://github.com/stijnvanhulle/template/commit/e0639c2936bad332a9fd7d95aa80bca796a85bab) Thanks [@stijnvanhulle](https://github.com/stijnvanhulle)! - Fold the `ponytail` skill and its `/ponytail` command into `deslop`, so one skill and one + `/deslop` command run both checks in a single pass. + + - Runs the reuse-first decision ladder (over-engineering: unneeded deps, wrappers, config) + alongside the existing AI style-tell checklist (needless comments, defensive checks, `any` + casts). + - Lists every finding with the file, what it violates, and the proposed fix, then asks before + changing anything, using a native picker where the client has one and a lettered list + otherwise. `deslop` used to apply its fixes directly; it now asks first, matching how + `ponytail` already worked. + - Applies only the findings you confirm. + + To upgrade, run `/deslop` where you used to run `/ponytail`. + +### Minor Changes + +- [#226](https://github.com/stijnvanhulle/template/pull/226) [`e0639c2`](https://github.com/stijnvanhulle/template/commit/e0639c2936bad332a9fd7d95aa80bca796a85bab) Thanks [@stijnvanhulle](https://github.com/stijnvanhulle)! - The `user-questions` rule now says to always reach for the native picker (`AskUserQuestion` or + a client's equivalent) when a task is blocked on information only the user can supply, instead + of reserving it for a fixed-choice fork. + + - Covers open-ended requests too: offer your best-guess options and let `Other` carry the real + answer, rather than falling back to a plain question. + - The one carve-out is a secret or credential (an API key, a token, a password), which stays a + plain question and is never echoed back or offered as a guessable option. + +- [#226](https://github.com/stijnvanhulle/template/pull/226) [`e0639c2`](https://github.com/stijnvanhulle/template/commit/e0639c2936bad332a9fd7d95aa80bca796a85bab) Thanks [@stijnvanhulle](https://github.com/stijnvanhulle)! - `code-reviewer` now folds `deslop`'s reuse-first ladder and AI style tells, `jsdoc`'s + conventions, `humanizer`'s pattern list, and `documentation`'s style guide into its review, in + that order, then asks which findings to hand off for fixing. + + - Stays read-only: it reports findings by these skills' criteria instead of applying their + fixes, then follows the `user-questions` rule to ask which of `deslop`, `jsdoc`, `humanizer`, + or `documentation` you want run next. + - A pass with nothing in its category (no exported symbols touched, no prose changed) is + skipped rather than forced. + +- [#226](https://github.com/stijnvanhulle/template/pull/226) [`e0639c2`](https://github.com/stijnvanhulle/template/commit/e0639c2936bad332a9fd7d95aa80bca796a85bab) Thanks [@stijnvanhulle](https://github.com/stijnvanhulle)! - `deslop` now runs the `humanizer` pattern list as a third pass over any changed README, doc, + comment block, or other user-facing markdown in the diff, alongside its existing reuse-first + ladder and code style-tell checks. + + - One `/deslop` run now covers over-engineering, code style tells, and AI writing tells in + prose, instead of needing a separate `/humanizer` pass for prose changes. + - `humanizer` still works standalone for prose review outside a code diff. + +### Patch Changes + +- [#226](https://github.com/stijnvanhulle/template/pull/226) [`e0639c2`](https://github.com/stijnvanhulle/template/commit/e0639c2936bad332a9fd7d95aa80bca796a85bab) Thanks [@stijnvanhulle](https://github.com/stijnvanhulle)! - Cut the token cost of the `jsdoc` and `pr` skills without dropping any guidance. + + - Moved the `jsdoc` skill's full `@example` gallery and documentation patterns into + `references/examples.md`, so the default load only carries the tag tables, guidelines, and tag + order. + - Shortened the `jsdoc` skill's frontmatter description, since it stays in context in every + `` listing whether or not the skill loads. + - Tightened repetitive prose across the `pr` skill's steps, keeping every check and guardrail. + ## 0.6.0 ### Minor Changes diff --git a/tools/cursor/package.json b/tools/cursor/package.json index 9452d43..fb4fae6 100644 --- a/tools/cursor/package.json +++ b/tools/cursor/package.json @@ -1,6 +1,6 @@ { "name": "@stijnvanhulle/template-cursor-plugin", - "version": "0.6.0", + "version": "1.0.0", "private": true, "description": "Cursor plugin for this template (agents, commands, rules, skills).", "license": "MIT"