Pinmind: orchestrate Skillstate checkpoints

.codex-plugin/plugin.json

@@ -1,6 +1,6 @@
 {
   "name": "pinmind",
-  "version": "0.10.1",
+  "version": "0.11.0",
   "description": "Adaptive RU/EN task controller with deterministic routing, outcome contracts, and verified evidence.",
   "author": {
     "name": "Pinmind Project"

.grok-plugin/plugin.json

@@ -1,6 +1,6 @@
 {
   "name": "pinmind",
-  "version": "0.10.1",
+  "version": "0.11.0",
   "description": "Adaptive RU/EN task controller with deterministic routing, outcome contracts, and verified evidence.",
   "author": {
     "name": "Pinmind Project"

.node-version

@@ -1 +1 @@
-24.19.0
+24.20.0

CHANGELOG.md

@@ -6,6 +6,21 @@ The format follows [Keep a Changelog](https://keepachangelog.com/en/2.0.0/), and
 
 ## [Unreleased]
 
+## [0.11.0] - 2026-09-01
+
+### Added
+
+- A single `UserPromptSubmit` orchestration hook that keeps Pinmind as the sole
+  controller and selects passive Skillstate checkpoints only for durable work.
+- Regression coverage for short-task exclusion, durable-task admission,
+  confirmation boundaries, and Ponytail's code-only role.
+
+### Changed
+
+- Document the fixed hierarchy: Pinmind controls, Skillstate records bounded
+  checkpoints, and Ponytail simplifies code.
+- Advance the pinned release runtime to Node `24.20.0`.
+
 ## [0.10.1] - 2026-08-24
 
 ### Added

LANGUAGE_ROUTING.md

@@ -1,6 +1,6 @@
 # Language routing and activation evaluation
 
-Status: Phase A is included in the stable `0.6` line, following its experimental introduction in `0.5.0-experimental`. The current `0.10.1` source has a **457-case deterministic routing regression corpus**, **64 closed-schema language-evaluation cases** (32 development and 32 frozen release-gate cases), **28 host-activation fixtures**, and a separate phased-transfer regression for `CONTEXT_READY`. The local evaluator and release gate are implemented; fresh-host selection and end-task utility remain separate, unproven layers. The release split is a fixed regression gate, not a statistically independent benchmark. This document does not claim that ChatGPT, Codex, or Grok will activate Pinmind for every possible phrase.
+Status: Phase A is included in the stable `0.6` line, following its experimental introduction in `0.5.0-experimental`. The current `0.11.0` source has a **457-case deterministic routing regression corpus**, **64 closed-schema language-evaluation cases** (32 development and 32 frozen release-gate cases), **28 host-activation fixtures**, and a separate phased-transfer regression for `CONTEXT_READY`. The local evaluator and release gate are implemented; fresh-host selection and end-task utility remain separate, unproven layers. The release split is a fixed regression gate, not a statistically independent benchmark. This document does not claim that ChatGPT, Codex, or Grok will activate Pinmind for every possible phrase.
 
 ## Decision
 

PRIVACY.md

@@ -1,6 +1,6 @@
 # Privacy
 
-Pinmind `0.10.1` is a skills-only plugin. It provides local instructions, a bundled kernel, and offline AEP, language-routing, and A/B/C evaluation inputs; it has no account system, connector, MCP server, telemetry service, or project-operated network endpoint.
+Pinmind `0.11.0` is a local skill and prompt-routing hook. It provides local instructions, a bundled kernel, and offline AEP, language-routing, and A/B/C evaluation inputs; it has no account system, connector, MCP server, telemetry service, or project-operated network endpoint. The hook returns routing context only and does not persist the submitted prompt.
 
 Pinmind processes only the task context and files made available by the active ChatGPT or Codex host. Data handling by that host and by any separately enabled tool is governed by the corresponding provider and user configuration.
 

README.md

@@ -4,7 +4,7 @@
 
 Pinmind is a Russian-and-English workflow controller packaged as a Grok skill and a Codex App plugin. It classifies non-trivial work, including colloquial and lightly misspelled Russian, applies a process proportional to risk, composes specialist skills, and requires current evidence before calling a task complete.
 
-Current stable version: `0.10.1`.
+Current stable version: `0.11.0`.
 
 - GitHub repository marketplace: included in this repository.
 - Universal Plugins Directory: **not listed yet**. ChatGPT catalog steps apply only after OpenAI approval and a live listing check.
@@ -16,18 +16,18 @@ Current stable version: `0.10.1`.
 1. Add the pinned stable repository marketplace:
 
    ```bash
-   codex plugin marketplace add iammedved/Pinmind --ref v0.10.1
+   codex plugin marketplace add iammedved/Pinmind --ref v0.11.0
    ```
 
 2. Start Codex, run `/plugins`, select **Pinmind Project**, and install **Pinmind**.
 3. Start a new Codex session.
 4. Run `/skills` and confirm that `pinmind` is available.
 
-The repository marketplace is separate from OpenAI's universal Plugins Directory. Pinning `v0.10.1` selects this exact stable release; omit `--ref` only when you intentionally want the reviewed latest repository state.
+The repository marketplace is separate from OpenAI's universal Plugins Directory. Pinning `v0.11.0` selects this exact stable release; omit `--ref` only when you intentionally want the reviewed latest repository state.
 
 ### Recommended design-enabled installation order
 
-Pinmind `0.10.1` can deterministically select two maintained external design
+Pinmind `0.11.0` can deterministically select two maintained external design
 skills. They are not bundled with Pinmind and must be installed separately.
 For a fresh project, use this order:
 
@@ -57,7 +57,7 @@ For a fresh project, use this order:
 3. Install the reviewed Pinmind release:
 
    ```bash
-   codex plugin marketplace add iammedved/Pinmind --ref v0.10.1
+   codex plugin marketplace add iammedved/Pinmind --ref v0.11.0
    ```
 
    Open `/plugins`, install **Pinmind** from **Pinmind Project**, and start a
@@ -67,7 +67,7 @@ For a fresh project, use this order:
    `ui-ux-pro-max` are available.
 
 The historical `frontend-skill` is deprecated upstream and is not part of
-Pinmind `0.10.1` design routing. Do not install it for this workflow.
+Pinmind `0.11.0` design routing. Do not install it for this workflow.
 
 ### Codex CLI: upgrade or reinstall a reviewed revision
 
@@ -93,7 +93,7 @@ recovery path; it is not the supported public upgrade path.
 4. If Pinmind does not appear immediately, restart Codex.
 5. Run `/skills` and confirm that `pinmind` is available.
 
-No extra setup is required. Pinmind is currently skills-only: it needs no connector, external account, API key, or MCP server.
+Pinmind needs no connector, external account, API key, or MCP server. The bundled local hook only classifies the submitted prompt and adds routing context; it performs no task action itself.
 
 ### Grok Build CLI: install for every new chat
 
@@ -130,7 +130,7 @@ Pinmind is not currently available in the public Plugins Directory. After the li
 4. Start a new chat.
 5. Send: `@Pinmind Audit this repository without changing files. Determine the route first and report only verified findings.`
 
-No connector, external account, API key, or additional configuration is required for the current skills-only package.
+No connector, external account, API key, or additional configuration is required for the current local skill-and-hook package.
 
 Official OpenAI guidance: [install and use plugins](https://learn.chatgpt.com/docs/plugins), [package plugins and repository marketplaces](https://developers.openai.com/plugins/build/plugins), and [build or install skills](https://learn.chatgpt.com/docs/build-skills).
 
@@ -146,11 +146,11 @@ Official OpenAI guidance: [install and use plugins](https://learn.chatgpt.com/do
 
 Implicit selection is probabilistic. Explicit invocation is the reliable choice for critical work.
 
-### Design-skill routing in 0.10.1
+### Design-skill routing in 0.11.0
 
 Earlier Pinmind releases could compose specialist skills only through general
 instructions. That left the choice between overlapping design skills to the
-host or the current prompt. Version `0.10.1` adds a deterministic post-route
+host or the current prompt. Version `0.11.0` includes a deterministic post-route
 decision:
 
 - **Impeccable only** for an explicit Impeccable command or an end-to-end
@@ -167,10 +167,9 @@ Pinmind makes this choice with `design route` after its main process route.
 The specialist result never grants permission to edit, publish, deploy, use
 network services, or change Figma. Those boundaries still come from the user.
 
-`0.10.1` is a patch release because the existing main route, contracts,
-lifecycle, and authority model remain backward compatible. The release fixes
-ambiguous design-skill composition, documents the external installation
-requirements, and excludes the deprecated `frontend-skill`.
+`0.11.0` is a minor release because Pinmind now owns automatic orchestration.
+It decides whether a task needs passive Skillstate checkpoints while keeping
+Ponytail limited to code simplification.
 
 Pinmind `0.10.0` added journaled `abandon`, non-destructive `archive`, proportional debugging/TDD/architecture/review guidance, and a compact readability profile. Superpowers and dyslex.ai remain research inputs: they are not installed, imported, or made runtime dependencies.
 
@@ -264,7 +263,7 @@ publish or display Codex cachebuster build metadata.
 
 The stable public line starts at `0.6.0`. Backward-compatible fixes use patch
 versions such as `0.6.1`, `0.6.2`, `0.6.3`, `0.8.1`, `0.8.2`, `0.9.1`, and `0.10.1`; backward-compatible feature releases use a
-new minor such as `0.7.0`, `0.8.0`, `0.9.0`, or `0.10.0`. Existing experimental tags are immutable history and
+new minor such as `0.7.0`, `0.8.0`, `0.9.0`, `0.10.0`, or `0.11.0`. Existing experimental tags are immutable history and
 are never moved to newer commits.
 
 ## Validation
@@ -288,8 +287,8 @@ this is a review-visible tamper-evidence boundary, not a cryptographically
 independent benchmark.
 
 For an auditable inventory, Pinmind counts top-level `test(` declarations rather
-than quoting Node's runtime summary. The current manifest records 121 declarations
-across ten test files, plus fixture-case counts for routes, activation, AEP,
+than quoting Node's runtime summary. The current manifest records 124 declarations
+across eleven test files, plus fixture-case counts for routes, activation, AEP,
 parallel admission, and language evaluation. These are separate dimensions and
 are not presented as one inflated "test count."
 

ROADMAP.md

@@ -2,8 +2,8 @@
 
 ## Verified baseline
 
-The current stable baseline is `0.10.1`, published as tag `v0.10.1`. See
-[CHANGELOG.md](CHANGELOG.md). The `0.10.1` release
+The current stable baseline is `0.11.0`, published as tag `v0.11.0`. See
+[CHANGELOG.md](CHANGELOG.md). The `0.11.0` release
 provides deterministic post-activation routing, local run-state containment and
 recovery diagnostics, baseline/freshness evidence, read-only `final check`, and
 explicit `finalize`. It also includes the reviewed P0 routing contrasts, required
@@ -47,15 +47,15 @@ only unfinished, measurable work; it is not implementation authority.
 
 ## Now
 
-The `0.10.1` release keeps speech-act routing, a compact `SKILL.md`, and a split
+The `0.11.0` release keeps speech-act routing, a compact `SKILL.md`, and a split
 kernel. AEP, A/B/C experiments, and P2 stay out of chat-time runtime. Remaining
 work is comparable host-run A/B/C evidence, fresh-host activation evidence, and
 independent release-corpus governance, not another workflow controller.
 
-### Remaining boundaries after the 0.10.1 release
+### Remaining boundaries after the 0.11.0 release
 
 - Re-run two fresh implicit and two fresh explicit Codex CLI sessions from the
-  immutable installed `v0.10.1` tag. Record the host version, package tree hash,
+  immutable installed `v0.11.0` tag. Record the host version, package tree hash,
   selected cache path, first route, and whether routing preceded task tools.
 - Directly inspect the Pinmind card and composer entry in a fresh Codex App
   session. The manifest and cache can prove which tracked image is configured;

evals/release-manifest.json

@@ -1,13 +1,18 @@
 {
   "schemaVersion": 1,
   "algorithm": "sha256",
-  "nodeVersion": "24.19.0",
+  "nodeVersion": "24.20.0",
   "frozenInputs": [
     {
       "role": "router",
       "path": "skills/pinmind/scripts/lib/route.mjs",
       "sha256": "808643bbb66cf8458a239a7be66bce9ef2f91a2e75901520fcaeae5dcec0b79c"
     },
+    {
+      "role": "orchestration-hook",
+      "path": "hooks/pinmind-orchestrator.mjs",
+      "sha256": "c6d0806d6d94d719b41b0e1e923f66670396fa820ac24becdefe16b8d721479a"
+    },
     {
       "role": "design-specialist-router",
       "path": "skills/pinmind/scripts/lib/design-skills.mjs",
@@ -70,6 +75,14 @@
     }
   ],
   "commands": [
+    {
+      "id": "orchestration-hook-tests",
+      "argv": [
+        "node",
+        "--test",
+        "tests/orchestration-hook.test.mjs"
+      ]
+    },
     {
       "id": "kernel-tests",
       "argv": [
@@ -282,6 +295,10 @@
   ],
   "inventory": {
     "testFiles": [
+      {
+        "path": "tests/orchestration-hook.test.mjs",
+        "topLevelDeclarations": 3
+      },
       {
         "path": "tests/kernel.test.mjs",
         "topLevelDeclarations": 63

hooks/hooks.json

@@ -0,0 +1,16 @@
+{
+  "hooks": {
+    "UserPromptSubmit": [
+      {
+        "hooks": [
+          {
+            "type": "command",
+            "command": "node \"${PLUGIN_ROOT}/hooks/pinmind-orchestrator.mjs\"",
+            "timeout": 2,
+            "statusMessage": "Pinmind is routing the task"
+          }
+        ]
+      }
+    ]
+  }
+}

hooks/pinmind-orchestrator.mjs

@@ -0,0 +1,49 @@
+#!/usr/bin/env node
+import { fileURLToPath } from 'node:url';
+
+import { routeTask } from '../skills/pinmind/scripts/lib/route.mjs';
+
+// Persistent Pinmind runs currently exist only for software changes.
+const DURABLE_ROUTE = new Set(['software-change']);
+
+export function needsSkillstate(prompt, route) {
+  if (!DURABLE_ROUTE.has(route.route) || route.needsHumanConfirmation) return false;
+  const text = String(prompt).trim();
+  const stages = [...text.matchAll(/\b(?:audit|analy[sz]e|inspect|implement|fix|test|verify|refactor|review|document|build|update|migrate|debug)\b|(?:аудит|проанализ|проверь|исправ|реализ|добав|протест|рефактор|подготов|настрой|обнов|внедр|мигрир|исслед)/giu)].length;
+  return text.length >= 600 || stages >= 3 || route.executionSpan !== 'local' || route.risk === 'high' || /(?:многоэтап|долг(?:ая|ий|ую)|нескольк(?:о|их) этап|interruption|resume|сначала.+(?:потом|затем))/iu.test(text);
+}
+
+export function hookOutputForPrompt(prompt) {
+  if (typeof prompt !== 'string' || !prompt.trim()) return {};
+  const route = routeTask({ text: prompt });
+  if (route.route === 'simple' || route.route === 'operational') return {};
+  const durable = needsSkillstate(prompt, route);
+  const statePolicy = durable
+    ? 'This task needs durable state. After Pinmind initializes its run, use $skillstate-runtime only for passive `skillstate checkpoint` records at clean phase boundaries. Never run `skillstate run` or start another controller.'
+    : 'This task does not need Skillstate. Do not start or invoke it.';
+  return {
+    hookSpecificOutput: {
+      hookEventName: 'UserPromptSubmit',
+      additionalContext: `origin=pinmind. Pinmind is the sole workflow controller for this turn; load and follow $pinmind before task tools. ${statePolicy} Ponytail may simplify code, but it never controls routing, state, authority, or phase order.`,
+    },
+  };
+}
+
+function startHook() {
+  let input = '';
+  let finished = false;
+  function finish() {
+    if (finished) return;
+    finished = true;
+    try { process.stdout.write(JSON.stringify(hookOutputForPrompt(JSON.parse(input.replace(/^\uFEFF/, '')).prompt))); }
+    catch { process.stdout.write('{}'); }
+  }
+  process.stdin.on('data', chunk => { input += chunk; });
+  process.stdin.on('end', finish);
+  process.stdin.on('error', finish);
+  process.stdin.resume();
+  const timeout = setTimeout(finish, 1000);
+  process.stdin.once('end', () => clearTimeout(timeout));
+}
+
+if (process.argv[1] === fileURLToPath(import.meta.url)) startHook();

scripts/verify-release.mjs

@@ -48,6 +48,7 @@ const TEST_FILE_KEYS = new Set(['path', 'topLevelDeclarations']);
 
 export const EXPECTED_FROZEN_INPUTS = Object.freeze([
   ['router', 'skills/pinmind/scripts/lib/route.mjs'],
+  ['orchestration-hook', 'hooks/pinmind-orchestrator.mjs'],
   ['design-specialist-router', 'skills/pinmind/scripts/lib/design-skills.mjs'],
   ['mixed-decomposition', 'skills/pinmind/scripts/lib/decomposition.mjs'],
   ['goal-context', 'skills/pinmind/scripts/lib/goal-context.mjs'],
@@ -63,6 +64,7 @@ export const EXPECTED_FROZEN_INPUTS = Object.freeze([
 ]);
 
 export const EXPECTED_COMMANDS = Object.freeze([
+  Object.freeze({ id: 'orchestration-hook-tests', argv: Object.freeze(['node', '--test', 'tests/orchestration-hook.test.mjs']) }),
   Object.freeze({ id: 'kernel-tests', argv: Object.freeze(['node', '--test', 'tests/kernel.test.mjs']) }),
   Object.freeze({ id: 'terminal-lifecycle-tests', argv: Object.freeze(['node', '--test', 'tests/terminal-lifecycle.test.mjs']) }),
   Object.freeze({ id: 'goal-context-tests', argv: Object.freeze(['node', '--test', 'tests/goal-context.test.mjs']) }),
@@ -93,6 +95,7 @@ export const EXPECTED_COMMANDS = Object.freeze([
 ]);
 
 const EXPECTED_TEST_FILES = Object.freeze([
+  'tests/orchestration-hook.test.mjs',
   'tests/kernel.test.mjs',
   'tests/terminal-lifecycle.test.mjs',
   'tests/goal-context.test.mjs',

skills/pinmind/SKILL.md

@@ -54,6 +54,19 @@ Check `.pinmind/active.json` before a new persistent run. Reconcile it read-only
 
 Keep specialist skills (PDF, Superpowers, `/design`, `/execute-plan`). Pinmind wraps them; it does not replace them.
 
+## Durable state and code simplicity
+
+Pinmind is the only workflow controller. The bundled `UserPromptSubmit` hook may mark a non-trivial request with `origin=pinmind` and decide whether it is durable enough for Skillstate. Never let Skillstate route the task, choose actions, or start its `run` controller beneath Pinmind.
+
+When the hook selects durable state, keep `.pinmind` as the source of truth. After `init` and at clean contract, execution, verification, and final phase boundaries, run:
+
+```bash
+skillstate doctor --workspace <path>
+skillstate checkpoint --workspace <path> --pinmind-run <run-id>
+```
+
+If either command fails, continue under Pinmind and report that durable checkpoints are unavailable; do not start a second controller. Skip Skillstate for short or one-step work. Ponytail may select the smallest code change and review over-engineering, but it never changes routing, authority, state, or phase order.
+
 ## UI design specialists
 
 After the main route, if the request may design, change, review, or fix a visual interface, run the deterministic specialist router before loading a UI skill:

tests/kernel.test.mjs

@@ -551,15 +551,15 @@ test('public release documentation, license, metadata, evaluation guides, and he
   const escapedBaseVersion = baseVersion.replace(/[.*+?^${}()|[\]\\]/g, '\\$&');
 
   assert.match(manifest.version, /^\d+\.\d+\.\d+$/);
-  assert.equal(baseVersion, '0.10.1');
+  assert.equal(baseVersion, '0.11.0');
   assert.equal(grokPlugin.name, manifest.name);
   assert.equal(grokPlugin.version, baseVersion);
   assert.match(manifest.description, /^Adaptive RU\/EN task controller/);
   assert.match(description, /^"Default RU\/EN controller/);
   assert.match(agent, /short_description:\s*"Adaptive verified RU\/EN task controller"/);
   for (const section of ['## Install, configure, and run', '## What Pinmind does', '## Kernel CLI', '## Versioning', '## Limitations']) assert.match(readme, new RegExp(section));
   assert.match(readme, new RegExp(`Current stable version:\\s*\`${escapedBaseVersion}\``));
-  assert.match(readme, /codex plugin marketplace add iammedved\/Pinmind --ref v0\.10\.1/);
+  assert.match(readme, /codex plugin marketplace add iammedved\/Pinmind --ref v0\.11\.0/);
   assert.doesNotMatch(readme, /--ref v0\.9\.1/);
   assert.doesNotMatch(readme, /Current (?:source|stable) version:[^\n]*unreleased/i);
   assert.match(readme, /Universal Plugins Directory:\s*\*\*not listed yet\*\*/);