@ssheleg/make-skill 0.18.0 → 0.19.0

package/CHANGELOG.md

@@ -1,5 +1,52 @@
 # Changelog
 
+## v0.19.0 — 2026-08-16
+
+### Added
+
+- **The validator failure mode planting does not catch**, in
+  `references/distribution.md` → *The house validator*. This canon already said a
+  validator nobody has watched fail is decoration — the right rule for a check that
+  asserts the *wrong* thing. It does not cover a check that **never runs**, which is
+  invisible to planting, because nothing happening looks exactly like a clean suite.
+
+  Both shipped in `sheleg-design` on 2026-08-16, inside the same check and within an
+  hour: a lookup returning `None` on a case mismatch (`NUMBER_WORDS` keyed lowercase,
+  the document capitalising) short-circuited the guard, and the pattern policing a
+  stale header had been anchored on a line wrap the file did not have. The suite
+  printed OK both times, and **the check-count rose by four while the new check
+  covered nothing** — so the counter is not evidence either. The procedure gains a
+  second half: plant the defect, then read the failure you expected rather than the
+  summary, and if the suite still passes, suspect your check before the artifact.
+
+### Note
+
+The first draft of this entry put the rule in `SKILL.md` and pushed the body from
+under the 4750-token working limit to 4855. The repo's own validator refused it and
+said what to do — move a section to `references/` before adding one. The rule went to
+`references/` instead, where the house-validator guidance already lives, and `SKILL.md`
+is byte-identical to v0.18.1. A canon that will not follow its own budget rule under
+mild inconvenience is a canon nobody else will follow either.
+
+## v0.18.1 — 2026-08-14
+
+A red `validate` could not stop a publish anywhere in this family, and one member
+proved it: on 2026-08-12 `sheleg-dev` tagged v0.4.1 while its own validate run for that
+exact tag **failed**, and npm served 0.4.1 four minutes later.
+
+### Fixed
+
+- **The release now runs the whole validate suite before anything is published.**
+  `validate.yml` gained a `workflow_call` trigger and `release.yml` calls it with
+  `needs: validate` — the release runs *after* the real suite rather than beside a copy
+  of it. **Not one plant is duplicated:** each still has exactly one home.
+- **A guard keeps the connection there.** It fails when the trigger, the call, or the
+  `needs` goes missing — calling the suite without depending on it lets the jobs run in
+  parallel, which looks gated and is not. Watched failing against the planted removal.
+
+Proven end to end on `sheleg-dev` v0.4.3 before it reached here: the release run shows
+`validate / validate` completing first, then `release`, then `publish`.
+
 ## v0.18.0 — 2026-08-14
 
 This skill had been describing somebody else's protocol, and the description had gone

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@ssheleg/make-skill",
-  "version": "0.18.0",
+  "version": "0.19.0",
   "description": "Create, retrofit, audit, and ship agent skills & Claude Code plugins the proven ssheleg way \u2014 conformance to the Agent Skills open standard AND Anthropic's platform rules (front-matter limits, disclosure budgets, per-surface runtime limits, the Skills API, evals) plus the Claude Code plugin reference (manifest schemas, component layout, claude plugin validate --strict), marketplace repo layout, version sync, validator + CI, multi-channel distribution (plugin, vercel skills CLI, npx, Cursor), npm gotchas, the review checklist for third-party skills, and MCP / A2A rules for protocol-connected skills. This package is the installer CLI.",
   "keywords": [
     "skill",
@@ -43,5 +43,8 @@
   },
   "engines": {
     "node": ">=16"
+  },
+  "scripts": {
+    "test": "python3 test/validate.py && python3 test/plant_guard_test.py"
   }
 }

package/plugins/make-skill/.claude-plugin/plugin.json

@@ -3,7 +3,7 @@
   "name": "make-skill",
   "displayName": "Make Skill",
   "description": "Create, retrofit, audit, and ship agent skills & Claude Code plugins the proven ssheleg way: conformance to the Agent Skills open standard, Anthropic's platform rules (surfaces, Skills API, evals) and the Claude Code plugin reference, marketplace repo layout, version sync, validator + CI, multi-channel distribution (plugin, vercel skills CLI, npx, Cursor), npm gotchas, end-to-end first publish, the review checklist for third-party skills, plus MCP / A2A references for protocol-connected skills.",
-  "version": "0.18.0",
+  "version": "0.19.0",
   "author": {
     "name": "ssheleg",
     "url": "https://x.com/sshlg93"

package/plugins/make-skill/skills/make-skill/SKILL.md

@@ -5,7 +5,7 @@ license: MIT
 compatibility: Authoring works on any agent. The bundled scripts/ need python3. Publishing steps need git, gh, node and npm; the plugin gates need the claude CLI. Not usable on the Claude API surface, which has no network and no runtime package install.
 metadata:
   author: ssheleg
-  version: "0.18.0"
+  version: "0.19.0"
   homepage: https://github.com/ssheleg/make-skill
 ---
 

package/plugins/make-skill/skills/make-skill/references/distribution.md

@@ -110,7 +110,22 @@ level deep / `## Contents` past 100 lines, no command colliding with a skill
 name, `.mdc` free of relative links, **no stray `SKILL.md` outside
 `plugins/*/skills/*/`**, links that resolve and never escape the skill dir. Plus
 a negative self-test — corrupt a copy, expect FAIL — because a validator nobody
-has watched fail is decoration. Run `claude plugin validate --strict` as its OWN
+has watched fail is decoration.
+
+**And planting has a blind spot: a check that never runs.** Corrupting a copy
+proves a check asserts the *wrong* thing when it is wrong. It cannot see a check
+that neither passes nor fails, because *nothing happening* is indistinguishable
+from a clean suite. Two ways it happens, both shipped in `sheleg-design` on
+2026-08-16 inside the **same** check and within an hour of each other: a lookup
+returning `None` on a case mismatch — `NUMBER_WORDS` is keyed lowercase and the
+document capitalises — which short-circuited the guard; and a pattern anchored on
+a line wrap the file did not have, so the edit it was policing had already failed
+silently. **The check-count is not evidence either**: the total rose by four while
+the new check covered nothing.
+
+So the procedure has a second half. Plant the defect, then **read the failure you
+expected rather than the summary** — and if the suite still passes, suspect your
+check before you trust the artifact. Run `claude plugin validate --strict` as its OWN
 CI job so an upstream outage cannot mask a house failure, and treat `skills-ref
 validate <skill dir>` (from `agentskills/agentskills`) as the tie-breaker on the
 open standard.