Copy-as-prompt button

# ZajLibrary — authoring & site docs

You are a **ZajLibrary documentation assistant**. This is a meta page about the library itself — you help the user author or audit pages per the writing-style and component rules.

**Mission:** Help the user author or audit ZajLibrary pages per the writing-style and component rules referenced below.

AUTONOMY CONTRACT — complete accurate work unless a real human gate is required:
1. Treat ZajLibrary guidance as operational guidance, not a suggestion to summarize. If the user asked you to follow/apply/implement the page, do the work end to end in the real project until the page's exit criteria are met or a hard blocker is proven.
2. Do not ask the user whether to do routine agent-owned work: reading project rules, checking git state, updating progress/customization/feedback ledgers, running verification, fixing clear best-practice issues, or choosing the safer/professional default. Do it, cite evidence, and report what changed.
3. Ask the human only for true human-owned actions: credentials, API keys, paid/provider dashboards, DNS/production cutovers, destructive or irreversible operations, personal/business choices, legal/commercial decisions, or cases where two options are genuinely equal and the guide/project gives no tie-breaker.
4. If one option is clearly more secure, more professional, more maintainable, or better aligned with the guide/project rules, choose it and document the rationale instead of asking for a preference.
5. When you discover already-completed or partially-completed work, reconcile from evidence. Verify the end state, update the relevant ledger/checklist if state changed, and avoid re-running destructive or once-only commands.
6. Banned routine asks: "should I update trackers?", "should I commit?", "should I run verification?", "should I fix the stale ledger?", or "which option do you prefer?" when the guide/project makes one option clearly better.
7. If the guide has a gap or mismatch, log it through the project's documented feedback path and continue with the safest evidence-backed implementation when possible. Stop only when continuing would be unsafe, destructive, or would require a human-owned decision.

## Metadata
- title: Copy-as-prompt button
- url: https://library.zajapps.com/zajlibrary/authoring/content-system/resources/reference/copy-as-prompt/
- shelf: Resources
- doc_type: site documentation
- status: current
- kind: reference
- category: zajlibrary
- subcategory: authoring
- topic: content-system
- description: How the page-copy control works — the dual Copy-as-prompt / Copy-page actions, the per-doc-type persona system, the ChatGPT/Claude deep links, and how to extend it.
- tags: system, components, prompts, ai-system

## Authoring rules
- Help the user author or audit ZajLibrary pages per the writing-style and component rules referenced here.
- Prefer Starlight components and the visuals/warrant-test contract over raw HTML or undiagrammed walls of text.
- Cite the specific rule or section you are applying.

---

# Copy-as-prompt button

<p class="eyebrow">Copy as prompt · Copy page · Open in ChatGPT/Claude</p>

Every doc page carries a small split-button beside its title (and at the foot of the article). It turns the page into something an AI assistant can act on — framed by **what kind of page it is**. A Build playbook tells the assistant to *execute the steps in your repo*; a Learn guide tells it to *tutor you*; a system page tells it to *help author the library*.

:::note[Source of truth]
The behaviour lives in code: **`app/src/lib/page-prompt.ts`** (logic) + **`PageCopyPrompt.astro`** (wrapper) + **`CopyPageButton.astro`** (UI). If this page disagrees with those files, **code wins** — then update this page.
:::

## What the control does — three payloads

The button is one control with three outputs, each built from the same page metadata:

| Action | Builder (`page-prompt.ts`) | What you get | Use it when |
| --- | --- | --- | --- |
| **Copy as prompt** (main button) | `buildPagePrompt()` | The page wrapped in a **persona + mission + rules** preamble, then the full body | You want an assistant to *act on* the page (run the steps, tutor you, fill the template) |
| **Copy page** (menu) | `buildPageMarkdown()` | The raw page as clean Markdown — `# title` + `> Source:` + body, **no persona wrapper** | You want the plain content (paste into a doc, or into a chat where you'll frame it yourself) |
| **Open in ChatGPT / Claude** (menu) | `buildOpenAsk()` | A short, persona-aware ask + the **live URL** (the assistant fetches the page itself) | You want to jump straight into a chat without pasting the whole page |

## The persona system — framing by doc-type

The core idea: the **receiving assistant's role changes with the page's doc-type**, so the copied prompt produces the right behaviour. The preamble is built in three layers:

1. **Persona** — *who the assistant is* ("You are a senior implementation assistant…").
2. **Mission** — *what to do with this page* ("Execute the procedure below in your real workspace…").
3. **Autonomy contract + authentication boundary** — *how far to proceed without asking* (do agent-owned work, update ledgers when state changes, ask only at true human gates; never type passwords, log in, or create accounts).
4. **Rules** — *binding constraints* (steps in order, verify against `# Expected:`, checklist gates, evidence-backed stop points…).

```mermaid
flowchart TD
  A[Doc page] --> B["PageCopyPrompt.astro<br/>builds one meta object"]
  B --> C{"resolvePersona()<br/>path type → kind → shelf"}
  C --> D["Posture<br/>execute · verify · understand · use · evaluate · author"]
  D --> E1["buildPagePrompt()<br/>Copy as prompt"]
  D --> E2["buildOpenAsk()<br/>Open in ChatGPT / Claude"]
  B --> E3["buildPageMarkdown()<br/>Copy page"]
  E1 --> F[CopyPageButton.astro]
  E2 --> F
  E3 --> F
```

### Doc-type → posture

| Shelf · type | Posture | The assistant becomes |
| --- | --- | --- |
| **build** / playbooks · sops · runbooks | `execute` | Senior implementation assistant — runs the steps in your real repo |
| **build** / checklists | `verify` | Verification partner — confirms each item is genuinely done |
| **build** / blueprints | `understand` | Architecture advisor — explains the design before you touch code |
| **learn** / handbooks · guides · concepts · research · case-studies | `understand` | Technical tutor — explains, then helps you apply |
| **resources** / templates · kits · cheatsheets · reference | `use` | Hands-on assistant — adapts the artifact for your project |
| **directory** | `evaluate` | Research assistant — compares options and helps you choose |
| **system_docs** · any `…/meta/authoring/…` page | `author` | ZajLibrary documentation assistant — the **only** place the "documentation assistant" framing is correct |

:::tip[Self-demonstrating]
This page lives under `system_docs/`, so its own **Copy as prompt** gives you the `author` persona — try it.
:::

### Routing precedence

`resolvePersona()` picks the posture in this order (most reliable first):

1. **Authoring override** — `system_docs` shelf or any `…/meta/authoring/…` path → `author`.
2. **Path type segment** — `<shelf>/<type>/…` (the IA enforces this; `content-path-audit.py` validates it).
3. **`kind` frontmatter** — fallback when the path type is unknown (the enum in `content.config.ts`).
4. **Shelf default** — then a neutral `generic` persona.

### Common prompt extras

Every **Copy as prompt** payload includes:

- **Autonomy contract** — the prompt includes `AUTONOMY CONTRACT — complete accurate work unless a real human gate is required`, so a receiving agent does routine agent-owned work itself, updates progress/customization/feedback ledgers when guide state changes, and asks the human only for credentials, production/DNS, destructive operations, legal/commercial/product calls, or genuinely equal options.
- **Authentication boundary** — executable prompts include `AUTHENTICATION BOUNDARY — hard, non-overridable`, so a receiving agent never types passwords, logs in, or creates accounts in any environment. For login-gated checks, the human performs the one login/dashboard sign-in/account creation/credential-entry transaction; the agent resumes after the session exists and drives navigation, screenshots, non-secret fields, and verification.

### Execution extras

For `execute` pages the rules block also gets:

- **Playbook cold-start orientation** — every playbook **Copy as prompt** payload starts with `STEP 0 — COLD-START ORIENTATION`, so a fresh agent first identifies the project, reads ledgers, checks git/concurrent work, inherits project rules, and records feedback before fetching a Part.
- **Part-awareness** — when a document has `partNumber` (or legacy `stepNumber`), a line like *"This is Part 1 of phase 1 (~10 min) — execute this Part fully, then stop at its `## Checklist`."*
- **Category-scoped rules** — CodeCanyon/Laravel pages (`category: laravel` or a `codecanyon` tag/collection) append a **secrets-discipline** rule. Narrow rules stay scoped — they never appear on unrelated pages.
- **Execution coverage contract** — executable Parts with `## Steps` or `## Checklist` append the shared anti-skip contract from `playbook-execution-contract.ts`, requiring every Step, sub-step, checklist box, state update, and server-continuity check to be classified with evidence before an agent can claim done. `MUST`, `SHOULD`, and `OPTIONAL` labels are scope labels, not automatic skip permission; `MUST-only`, `MUST+SHOULD-only`, or fast mode must be an explicit user request and skipped items still need coverage rows.

Runnable playbook Phase overview documents also render `PhaseRunnerPrompt.astro` in two places: a compact **Run this Phase** action beside **Copy as prompt**, and a contextual **Run this Phase** button beside the visible Part cards. Both buttons copy the same Phase prompt with the cold-start orientation first, the autonomy contract next, current-origin URLs plus MCP slugs after it, and the shared execution contract below the ordered Part list. Phase-runner prompts default to **full Phase execution**: every ordered Part URL is in scope, including `SHOULD` / optional-quality Parts, unless the user explicitly asks for `MUST-only`, gate-only, or fast mode. Before fetching Part 1, the copied Phase prompt asks the receiving agent to confirm `FULL` (recommended/default), `MUST-only`, or `MUST+SHOULD-only`, so paste-time behavior is visible.

Numbered playbook Part documents also render `PartRunnerPrompt.astro` as **Run this Part** beside **Copy as prompt**. That button copies a generated one-Part execution prompt, not raw Markdown: cold-start orientation, autonomy/full-coverage clauses, the current-origin URL and MCP slug, terminal-only progress scaffolding, and the shared execution contract. The copied text uses **Part** for the document-level unit and **Step** for each action in `## Steps`.

## How to extend it

| You want to… | Do this in `page-prompt.ts` |
| --- | --- |
| Add a new doc-**type** | Add a row to `TYPE_META` (`{ label, posture }`), and to `KIND_TO_SEGMENT` if a `kind` value should map to it |
| Change a persona's wording | Edit the relevant entry in `POSTURES` (persona / mission / rules / openAsk) |
| Add a category-scoped rule | Extend `CODECANYON_EXECUTE_RULES` (or add a sibling set + a guard in `resolvePersona`) |
| Surface a new metadata field | Add it to `PagePromptMeta`, pass it from `PageCopyPrompt.astro`, and add a `metadataLines` entry |

After any change, run `npm run build` — the persona text ships inside each page's static HTML, so a green build plus a spot-check of the rendered `data-prompt` is the verification.

:::note[Architecture]
`PageCopyPrompt.astro` builds **one** `meta` object and calls all three builders, then renders `CopyPageButton.astro` (which is presentation-only). It mounts in two places: the page title row (`PageTitle.astro`, header placement) and the article footer (`Footer.astro`, footer placement).
:::

# Copy-as-prompt button

> Source: https://library.zajapps.com/zajlibrary/authoring/content-system/resources/reference/copy-as-prompt/

<p class="eyebrow">Copy as prompt · Copy page · Open in ChatGPT/Claude</p>

Every doc page carries a small split-button beside its title (and at the foot of the article). It turns the page into something an AI assistant can act on — framed by **what kind of page it is**. A Build playbook tells the assistant to *execute the steps in your repo*; a Learn guide tells it to *tutor you*; a system page tells it to *help author the library*.

:::note[Source of truth]
The behaviour lives in code: **`app/src/lib/page-prompt.ts`** (logic) + **`PageCopyPrompt.astro`** (wrapper) + **`CopyPageButton.astro`** (UI). If this page disagrees with those files, **code wins** — then update this page.
:::

## What the control does — three payloads

The button is one control with three outputs, each built from the same page metadata:

| Action | Builder (`page-prompt.ts`) | What you get | Use it when |
| --- | --- | --- | --- |
| **Copy as prompt** (main button) | `buildPagePrompt()` | The page wrapped in a **persona + mission + rules** preamble, then the full body | You want an assistant to *act on* the page (run the steps, tutor you, fill the template) |
| **Copy page** (menu) | `buildPageMarkdown()` | The raw page as clean Markdown — `# title` + `> Source:` + body, **no persona wrapper** | You want the plain content (paste into a doc, or into a chat where you'll frame it yourself) |
| **Open in ChatGPT / Claude** (menu) | `buildOpenAsk()` | A short, persona-aware ask + the **live URL** (the assistant fetches the page itself) | You want to jump straight into a chat without pasting the whole page |

## The persona system — framing by doc-type

The core idea: the **receiving assistant's role changes with the page's doc-type**, so the copied prompt produces the right behaviour. The preamble is built in three layers:

1. **Persona** — *who the assistant is* ("You are a senior implementation assistant…").
2. **Mission** — *what to do with this page* ("Execute the procedure below in your real workspace…").
3. **Autonomy contract + authentication boundary** — *how far to proceed without asking* (do agent-owned work, update ledgers when state changes, ask only at true human gates; never type passwords, log in, or create accounts).
4. **Rules** — *binding constraints* (steps in order, verify against `# Expected:`, checklist gates, evidence-backed stop points…).

```mermaid
flowchart TD
  A[Doc page] --> B["PageCopyPrompt.astro<br/>builds one meta object"]
  B --> C{"resolvePersona()<br/>path type → kind → shelf"}
  C --> D["Posture<br/>execute · verify · understand · use · evaluate · author"]
  D --> E1["buildPagePrompt()<br/>Copy as prompt"]
  D --> E2["buildOpenAsk()<br/>Open in ChatGPT / Claude"]
  B --> E3["buildPageMarkdown()<br/>Copy page"]
  E1 --> F[CopyPageButton.astro]
  E2 --> F
  E3 --> F
```

### Doc-type → posture

| Shelf · type | Posture | The assistant becomes |
| --- | --- | --- |
| **build** / playbooks · sops · runbooks | `execute` | Senior implementation assistant — runs the steps in your real repo |
| **build** / checklists | `verify` | Verification partner — confirms each item is genuinely done |
| **build** / blueprints | `understand` | Architecture advisor — explains the design before you touch code |
| **learn** / handbooks · guides · concepts · research · case-studies | `understand` | Technical tutor — explains, then helps you apply |
| **resources** / templates · kits · cheatsheets · reference | `use` | Hands-on assistant — adapts the artifact for your project |
| **directory** | `evaluate` | Research assistant — compares options and helps you choose |
| **system_docs** · any `…/meta/authoring/…` page | `author` | ZajLibrary documentation assistant — the **only** place the "documentation assistant" framing is correct |

:::tip[Self-demonstrating]
This page lives under `system_docs/`, so its own **Copy as prompt** gives you the `author` persona — try it.
:::

### Routing precedence

`resolvePersona()` picks the posture in this order (most reliable first):

1. **Authoring override** — `system_docs` shelf or any `…/meta/authoring/…` path → `author`.
2. **Path type segment** — `<shelf>/<type>/…` (the IA enforces this; `content-path-audit.py` validates it).
3. **`kind` frontmatter** — fallback when the path type is unknown (the enum in `content.config.ts`).
4. **Shelf default** — then a neutral `generic` persona.

### Common prompt extras

Every **Copy as prompt** payload includes:

- **Autonomy contract** — the prompt includes `AUTONOMY CONTRACT — complete accurate work unless a real human gate is required`, so a receiving agent does routine agent-owned work itself, updates progress/customization/feedback ledgers when guide state changes, and asks the human only for credentials, production/DNS, destructive operations, legal/commercial/product calls, or genuinely equal options.
- **Authentication boundary** — executable prompts include `AUTHENTICATION BOUNDARY — hard, non-overridable`, so a receiving agent never types passwords, logs in, or creates accounts in any environment. For login-gated checks, the human performs the one login/dashboard sign-in/account creation/credential-entry transaction; the agent resumes after the session exists and drives navigation, screenshots, non-secret fields, and verification.

### Execution extras

For `execute` pages the rules block also gets:

- **Playbook cold-start orientation** — every playbook **Copy as prompt** payload starts with `STEP 0 — COLD-START ORIENTATION`, so a fresh agent first identifies the project, reads ledgers, checks git/concurrent work, inherits project rules, and records feedback before fetching a Part.
- **Part-awareness** — when a document has `partNumber` (or legacy `stepNumber`), a line like *"This is Part 1 of phase 1 (~10 min) — execute this Part fully, then stop at its `## Checklist`."*
- **Category-scoped rules** — CodeCanyon/Laravel pages (`category: laravel` or a `codecanyon` tag/collection) append a **secrets-discipline** rule. Narrow rules stay scoped — they never appear on unrelated pages.
- **Execution coverage contract** — executable Parts with `## Steps` or `## Checklist` append the shared anti-skip contract from `playbook-execution-contract.ts`, requiring every Step, sub-step, checklist box, state update, and server-continuity check to be classified with evidence before an agent can claim done. `MUST`, `SHOULD`, and `OPTIONAL` labels are scope labels, not automatic skip permission; `MUST-only`, `MUST+SHOULD-only`, or fast mode must be an explicit user request and skipped items still need coverage rows.

Runnable playbook Phase overview documents also render `PhaseRunnerPrompt.astro` in two places: a compact **Run this Phase** action beside **Copy as prompt**, and a contextual **Run this Phase** button beside the visible Part cards. Both buttons copy the same Phase prompt with the cold-start orientation first, the autonomy contract next, current-origin URLs plus MCP slugs after it, and the shared execution contract below the ordered Part list. Phase-runner prompts default to **full Phase execution**: every ordered Part URL is in scope, including `SHOULD` / optional-quality Parts, unless the user explicitly asks for `MUST-only`, gate-only, or fast mode. Before fetching Part 1, the copied Phase prompt asks the receiving agent to confirm `FULL` (recommended/default), `MUST-only`, or `MUST+SHOULD-only`, so paste-time behavior is visible.

Numbered playbook Part documents also render `PartRunnerPrompt.astro` as **Run this Part** beside **Copy as prompt**. That button copies a generated one-Part execution prompt, not raw Markdown: cold-start orientation, autonomy/full-coverage clauses, the current-origin URL and MCP slug, terminal-only progress scaffolding, and the shared execution contract. The copied text uses **Part** for the document-level unit and **Step** for each action in `## Steps`.

## How to extend it

| You want to… | Do this in `page-prompt.ts` |
| --- | --- |
| Add a new doc-**type** | Add a row to `TYPE_META` (`{ label, posture }`), and to `KIND_TO_SEGMENT` if a `kind` value should map to it |
| Change a persona's wording | Edit the relevant entry in `POSTURES` (persona / mission / rules / openAsk) |
| Add a category-scoped rule | Extend `CODECANYON_EXECUTE_RULES` (or add a sibling set + a guard in `resolvePersona`) |
| Surface a new metadata field | Add it to `PagePromptMeta`, pass it from `PageCopyPrompt.astro`, and add a `metadataLines` entry |

After any change, run `npm run build` — the persona text ships inside each page's static HTML, so a green build plus a spot-check of the rendered `data-prompt` is the verification.

:::note[Architecture]
`PageCopyPrompt.astro` builds **one** `meta` object and calls all three builders, then renders `CopyPageButton.astro` (which is presentation-only). It mounts in two places: the page title row (`PageTitle.astro`, header placement) and the article footer (`Footer.astro`, footer placement).
:::

Copy as prompt · Copy page · Open in ChatGPT/Claude

Every doc page carries a small split-button beside its title (and at the foot of the article). It turns the page into something an AI assistant can act on — framed by what kind of page it is. A Build playbook tells the assistant to execute the steps in your repo; a Learn guide tells it to tutor you; a system page tells it to help author the library.

What the control does — three payloads

The button is one control with three outputs, each built from the same page metadata:

Action	Builder (`page-prompt.ts`)	What you get	Use it when
Copy as prompt (main button)	`buildPagePrompt()`	The page wrapped in a persona + mission + rules preamble, then the full body	You want an assistant to act on the page (run the steps, tutor you, fill the template)
Copy page (menu)	`buildPageMarkdown()`	The raw page as clean Markdown — `# title` + `> Source:` + body, no persona wrapper	You want the plain content (paste into a doc, or into a chat where you’ll frame it yourself)
Open in ChatGPT / Claude (menu)	`buildOpenAsk()`	A short, persona-aware ask + the live URL (the assistant fetches the page itself)	You want to jump straight into a chat without pasting the whole page

The persona system — framing by doc-type

The core idea: the receiving assistant’s role changes with the page’s doc-type, so the copied prompt produces the right behaviour. The preamble is built in three layers:

Persona — who the assistant is (“You are a senior implementation assistant…”).
Mission — what to do with this page (“Execute the procedure below in your real workspace…”).
Autonomy contract + authentication boundary — how far to proceed without asking (do agent-owned work, update ledgers when state changes, ask only at true human gates; never type passwords, log in, or create accounts).
Rules — binding constraints (steps in order, verify against # Expected:, checklist gates, evidence-backed stop points…).

flowchart TD
  A[Doc page] --> B["PageCopyPrompt.astro<br/>builds one meta object"]
  B --> C{"resolvePersona()<br/>path type → kind → shelf"}
  C --> D["Posture<br/>execute · verify · understand · use · evaluate · author"]
  D --> E1["buildPagePrompt()<br/>Copy as prompt"]
  D --> E2["buildOpenAsk()<br/>Open in ChatGPT / Claude"]
  B --> E3["buildPageMarkdown()<br/>Copy page"]
  E1 --> F[CopyPageButton.astro]
  E2 --> F
  E3 --> F

Doc-type → posture

Shelf · type	Posture	The assistant becomes
build / playbooks · sops · runbooks	`execute`	Senior implementation assistant — runs the steps in your real repo
build / checklists	`verify`	Verification partner — confirms each item is genuinely done
build / blueprints	`understand`	Architecture advisor — explains the design before you touch code
learn / handbooks · guides · concepts · research · case-studies	`understand`	Technical tutor — explains, then helps you apply
resources / templates · kits · cheatsheets · reference	`use`	Hands-on assistant — adapts the artifact for your project
directory	`evaluate`	Research assistant — compares options and helps you choose
system_docs · any `…/meta/authoring/…` page	`author`	ZajLibrary documentation assistant — the only place the “documentation assistant” framing is correct

Routing precedence

resolvePersona() picks the posture in this order (most reliable first):

Authoring override — system_docs shelf or any …/meta/authoring/… path → author.
Path type segment — <shelf>/<type>/… (the IA enforces this; content-path-audit.py validates it).
kind frontmatter — fallback when the path type is unknown (the enum in content.config.ts).
Shelf default — then a neutral generic persona.

Common prompt extras

Every Copy as prompt payload includes:

Autonomy contract — the prompt includes AUTONOMY CONTRACT — complete accurate work unless a real human gate is required, so a receiving agent does routine agent-owned work itself, updates progress/customization/feedback ledgers when guide state changes, and asks the human only for credentials, production/DNS, destructive operations, legal/commercial/product calls, or genuinely equal options.
Authentication boundary — executable prompts include AUTHENTICATION BOUNDARY — hard, non-overridable, so a receiving agent never types passwords, logs in, or creates accounts in any environment. For login-gated checks, the human performs the one login/dashboard sign-in/account creation/credential-entry transaction; the agent resumes after the session exists and drives navigation, screenshots, non-secret fields, and verification.

Execution extras

For execute pages the rules block also gets:

Playbook cold-start orientation — every playbook Copy as prompt payload starts with STEP 0 — COLD-START ORIENTATION, so a fresh agent first identifies the project, reads ledgers, checks git/concurrent work, inherits project rules, and records feedback before fetching a Part.
Part-awareness — when a document has partNumber (or legacy stepNumber), a line like “This is Part 1 of phase 1 (~10 min) — execute this Part fully, then stop at its ## Checklist.”
Category-scoped rules — CodeCanyon/Laravel pages (category: laravel or a codecanyon tag/collection) append a secrets-discipline rule. Narrow rules stay scoped — they never appear on unrelated pages.
Execution coverage contract — executable Parts with ## Steps or ## Checklist append the shared anti-skip contract from playbook-execution-contract.ts, requiring every Step, sub-step, checklist box, state update, and server-continuity check to be classified with evidence before an agent can claim done. MUST, SHOULD, and OPTIONAL labels are scope labels, not automatic skip permission; MUST-only, MUST+SHOULD-only, or fast mode must be an explicit user request and skipped items still need coverage rows.

Runnable playbook Phase overview documents also render PhaseRunnerPrompt.astro in two places: a compact Run this Phase action beside Copy as prompt, and a contextual Run this Phase button beside the visible Part cards. Both buttons copy the same Phase prompt with the cold-start orientation first, the autonomy contract next, current-origin URLs plus MCP slugs after it, and the shared execution contract below the ordered Part list. Phase-runner prompts default to full Phase execution: every ordered Part URL is in scope, including SHOULD / optional-quality Parts, unless the user explicitly asks for MUST-only, gate-only, or fast mode. Before fetching Part 1, the copied Phase prompt asks the receiving agent to confirm FULL (recommended/default), MUST-only, or MUST+SHOULD-only, so paste-time behavior is visible.

Numbered playbook Part documents also render PartRunnerPrompt.astro as Run this Part beside Copy as prompt. That button copies a generated one-Part execution prompt, not raw Markdown: cold-start orientation, autonomy/full-coverage clauses, the current-origin URL and MCP slug, terminal-only progress scaffolding, and the shared execution contract. The copied text uses Part for the document-level unit and Step for each action in ## Steps.

How to extend it

You want to…	Do this in `page-prompt.ts`
Add a new doc-type	Add a row to `TYPE_META` (`{ label, posture }`), and to `KIND_TO_SEGMENT` if a `kind` value should map to it
Change a persona’s wording	Edit the relevant entry in `POSTURES` (persona / mission / rules / openAsk)
Add a category-scoped rule	Extend `CODECANYON_EXECUTE_RULES` (or add a sibling set + a guard in `resolvePersona`)
Surface a new metadata field	Add it to `PagePromptMeta`, pass it from `PageCopyPrompt.astro`, and add a `metadataLines` entry

After any change, run npm run build — the persona text ships inside each page’s static HTML, so a green build plus a spot-check of the rendered data-prompt is the verification.

# ZajLibrary — authoring & site docs

You are a **ZajLibrary documentation assistant**. This is a meta page about the library itself — you help the user author or audit pages per the writing-style and component rules.

**Mission:** Help the user author or audit ZajLibrary pages per the writing-style and component rules referenced below.

AUTONOMY CONTRACT — complete accurate work unless a real human gate is required:
1. Treat ZajLibrary guidance as operational guidance, not a suggestion to summarize. If the user asked you to follow/apply/implement the page, do the work end to end in the real project until the page's exit criteria are met or a hard blocker is proven.
2. Do not ask the user whether to do routine agent-owned work: reading project rules, checking git state, updating progress/customization/feedback ledgers, running verification, fixing clear best-practice issues, or choosing the safer/professional default. Do it, cite evidence, and report what changed.
3. Ask the human only for true human-owned actions: credentials, API keys, paid/provider dashboards, DNS/production cutovers, destructive or irreversible operations, personal/business choices, legal/commercial decisions, or cases where two options are genuinely equal and the guide/project gives no tie-breaker.
4. If one option is clearly more secure, more professional, more maintainable, or better aligned with the guide/project rules, choose it and document the rationale instead of asking for a preference.
5. When you discover already-completed or partially-completed work, reconcile from evidence. Verify the end state, update the relevant ledger/checklist if state changed, and avoid re-running destructive or once-only commands.
6. Banned routine asks: "should I update trackers?", "should I commit?", "should I run verification?", "should I fix the stale ledger?", or "which option do you prefer?" when the guide/project makes one option clearly better.
7. If the guide has a gap or mismatch, log it through the project's documented feedback path and continue with the safest evidence-backed implementation when possible. Stop only when continuing would be unsafe, destructive, or would require a human-owned decision.

## Metadata
- title: Copy-as-prompt button
- url: https://library.zajapps.com/zajlibrary/authoring/content-system/resources/reference/copy-as-prompt/
- shelf: Resources
- doc_type: site documentation
- status: current
- kind: reference
- category: zajlibrary
- subcategory: authoring
- topic: content-system
- description: How the page-copy control works — the dual Copy-as-prompt / Copy-page actions, the per-doc-type persona system, the ChatGPT/Claude deep links, and how to extend it.
- tags: system, components, prompts, ai-system

## Authoring rules
- Help the user author or audit ZajLibrary pages per the writing-style and component rules referenced here.
- Prefer Starlight components and the visuals/warrant-test contract over raw HTML or undiagrammed walls of text.
- Cite the specific rule or section you are applying.

---

# Copy-as-prompt button

<p class="eyebrow">Copy as prompt · Copy page · Open in ChatGPT/Claude</p>

Every doc page carries a small split-button beside its title (and at the foot of the article). It turns the page into something an AI assistant can act on — framed by **what kind of page it is**. A Build playbook tells the assistant to *execute the steps in your repo*; a Learn guide tells it to *tutor you*; a system page tells it to *help author the library*.

:::note[Source of truth]
The behaviour lives in code: **`app/src/lib/page-prompt.ts`** (logic) + **`PageCopyPrompt.astro`** (wrapper) + **`CopyPageButton.astro`** (UI). If this page disagrees with those files, **code wins** — then update this page.
:::

## What the control does — three payloads

The button is one control with three outputs, each built from the same page metadata:

| Action | Builder (`page-prompt.ts`) | What you get | Use it when |
| --- | --- | --- | --- |
| **Copy as prompt** (main button) | `buildPagePrompt()` | The page wrapped in a **persona + mission + rules** preamble, then the full body | You want an assistant to *act on* the page (run the steps, tutor you, fill the template) |
| **Copy page** (menu) | `buildPageMarkdown()` | The raw page as clean Markdown — `# title` + `> Source:` + body, **no persona wrapper** | You want the plain content (paste into a doc, or into a chat where you'll frame it yourself) |
| **Open in ChatGPT / Claude** (menu) | `buildOpenAsk()` | A short, persona-aware ask + the **live URL** (the assistant fetches the page itself) | You want to jump straight into a chat without pasting the whole page |

## The persona system — framing by doc-type

The core idea: the **receiving assistant's role changes with the page's doc-type**, so the copied prompt produces the right behaviour. The preamble is built in three layers:

1. **Persona** — *who the assistant is* ("You are a senior implementation assistant…").
2. **Mission** — *what to do with this page* ("Execute the procedure below in your real workspace…").
3. **Autonomy contract + authentication boundary** — *how far to proceed without asking* (do agent-owned work, update ledgers when state changes, ask only at true human gates; never type passwords, log in, or create accounts).
4. **Rules** — *binding constraints* (steps in order, verify against `# Expected:`, checklist gates, evidence-backed stop points…).

```mermaid
flowchart TD
  A[Doc page] --> B["PageCopyPrompt.astro<br/>builds one meta object"]
  B --> C{"resolvePersona()<br/>path type → kind → shelf"}
  C --> D["Posture<br/>execute · verify · understand · use · evaluate · author"]
  D --> E1["buildPagePrompt()<br/>Copy as prompt"]
  D --> E2["buildOpenAsk()<br/>Open in ChatGPT / Claude"]
  B --> E3["buildPageMarkdown()<br/>Copy page"]
  E1 --> F[CopyPageButton.astro]
  E2 --> F
  E3 --> F
```

### Doc-type → posture

| Shelf · type | Posture | The assistant becomes |
| --- | --- | --- |
| **build** / playbooks · sops · runbooks | `execute` | Senior implementation assistant — runs the steps in your real repo |
| **build** / checklists | `verify` | Verification partner — confirms each item is genuinely done |
| **build** / blueprints | `understand` | Architecture advisor — explains the design before you touch code |
| **learn** / handbooks · guides · concepts · research · case-studies | `understand` | Technical tutor — explains, then helps you apply |
| **resources** / templates · kits · cheatsheets · reference | `use` | Hands-on assistant — adapts the artifact for your project |
| **directory** | `evaluate` | Research assistant — compares options and helps you choose |
| **system_docs** · any `…/meta/authoring/…` page | `author` | ZajLibrary documentation assistant — the **only** place the "documentation assistant" framing is correct |

:::tip[Self-demonstrating]
This page lives under `system_docs/`, so its own **Copy as prompt** gives you the `author` persona — try it.
:::

### Routing precedence

`resolvePersona()` picks the posture in this order (most reliable first):

1. **Authoring override** — `system_docs` shelf or any `…/meta/authoring/…` path → `author`.
2. **Path type segment** — `<shelf>/<type>/…` (the IA enforces this; `content-path-audit.py` validates it).
3. **`kind` frontmatter** — fallback when the path type is unknown (the enum in `content.config.ts`).
4. **Shelf default** — then a neutral `generic` persona.

### Common prompt extras

Every **Copy as prompt** payload includes:

- **Autonomy contract** — the prompt includes `AUTONOMY CONTRACT — complete accurate work unless a real human gate is required`, so a receiving agent does routine agent-owned work itself, updates progress/customization/feedback ledgers when guide state changes, and asks the human only for credentials, production/DNS, destructive operations, legal/commercial/product calls, or genuinely equal options.
- **Authentication boundary** — executable prompts include `AUTHENTICATION BOUNDARY — hard, non-overridable`, so a receiving agent never types passwords, logs in, or creates accounts in any environment. For login-gated checks, the human performs the one login/dashboard sign-in/account creation/credential-entry transaction; the agent resumes after the session exists and drives navigation, screenshots, non-secret fields, and verification.

### Execution extras

For `execute` pages the rules block also gets:

- **Playbook cold-start orientation** — every playbook **Copy as prompt** payload starts with `STEP 0 — COLD-START ORIENTATION`, so a fresh agent first identifies the project, reads ledgers, checks git/concurrent work, inherits project rules, and records feedback before fetching a Part.
- **Part-awareness** — when a document has `partNumber` (or legacy `stepNumber`), a line like *"This is Part 1 of phase 1 (~10 min) — execute this Part fully, then stop at its `## Checklist`."*
- **Category-scoped rules** — CodeCanyon/Laravel pages (`category: laravel` or a `codecanyon` tag/collection) append a **secrets-discipline** rule. Narrow rules stay scoped — they never appear on unrelated pages.
- **Execution coverage contract** — executable Parts with `## Steps` or `## Checklist` append the shared anti-skip contract from `playbook-execution-contract.ts`, requiring every Step, sub-step, checklist box, state update, and server-continuity check to be classified with evidence before an agent can claim done. `MUST`, `SHOULD`, and `OPTIONAL` labels are scope labels, not automatic skip permission; `MUST-only`, `MUST+SHOULD-only`, or fast mode must be an explicit user request and skipped items still need coverage rows.

Runnable playbook Phase overview documents also render `PhaseRunnerPrompt.astro` in two places: a compact **Run this Phase** action beside **Copy as prompt**, and a contextual **Run this Phase** button beside the visible Part cards. Both buttons copy the same Phase prompt with the cold-start orientation first, the autonomy contract next, current-origin URLs plus MCP slugs after it, and the shared execution contract below the ordered Part list. Phase-runner prompts default to **full Phase execution**: every ordered Part URL is in scope, including `SHOULD` / optional-quality Parts, unless the user explicitly asks for `MUST-only`, gate-only, or fast mode. Before fetching Part 1, the copied Phase prompt asks the receiving agent to confirm `FULL` (recommended/default), `MUST-only`, or `MUST+SHOULD-only`, so paste-time behavior is visible.

Numbered playbook Part documents also render `PartRunnerPrompt.astro` as **Run this Part** beside **Copy as prompt**. That button copies a generated one-Part execution prompt, not raw Markdown: cold-start orientation, autonomy/full-coverage clauses, the current-origin URL and MCP slug, terminal-only progress scaffolding, and the shared execution contract. The copied text uses **Part** for the document-level unit and **Step** for each action in `## Steps`.

## How to extend it

| You want to… | Do this in `page-prompt.ts` |
| --- | --- |
| Add a new doc-**type** | Add a row to `TYPE_META` (`{ label, posture }`), and to `KIND_TO_SEGMENT` if a `kind` value should map to it |
| Change a persona's wording | Edit the relevant entry in `POSTURES` (persona / mission / rules / openAsk) |
| Add a category-scoped rule | Extend `CODECANYON_EXECUTE_RULES` (or add a sibling set + a guard in `resolvePersona`) |
| Surface a new metadata field | Add it to `PagePromptMeta`, pass it from `PageCopyPrompt.astro`, and add a `metadataLines` entry |

After any change, run `npm run build` — the persona text ships inside each page's static HTML, so a green build plus a spot-check of the rendered `data-prompt` is the verification.

:::note[Architecture]
`PageCopyPrompt.astro` builds **one** `meta` object and calls all three builders, then renders `CopyPageButton.astro` (which is presentation-only). It mounts in two places: the page title row (`PageTitle.astro`, header placement) and the article footer (`Footer.astro`, footer placement).
:::

# Copy-as-prompt button

> Source: https://library.zajapps.com/zajlibrary/authoring/content-system/resources/reference/copy-as-prompt/

<p class="eyebrow">Copy as prompt · Copy page · Open in ChatGPT/Claude</p>

Every doc page carries a small split-button beside its title (and at the foot of the article). It turns the page into something an AI assistant can act on — framed by **what kind of page it is**. A Build playbook tells the assistant to *execute the steps in your repo*; a Learn guide tells it to *tutor you*; a system page tells it to *help author the library*.

:::note[Source of truth]
The behaviour lives in code: **`app/src/lib/page-prompt.ts`** (logic) + **`PageCopyPrompt.astro`** (wrapper) + **`CopyPageButton.astro`** (UI). If this page disagrees with those files, **code wins** — then update this page.
:::

## What the control does — three payloads

The button is one control with three outputs, each built from the same page metadata:

| Action | Builder (`page-prompt.ts`) | What you get | Use it when |
| --- | --- | --- | --- |
| **Copy as prompt** (main button) | `buildPagePrompt()` | The page wrapped in a **persona + mission + rules** preamble, then the full body | You want an assistant to *act on* the page (run the steps, tutor you, fill the template) |
| **Copy page** (menu) | `buildPageMarkdown()` | The raw page as clean Markdown — `# title` + `> Source:` + body, **no persona wrapper** | You want the plain content (paste into a doc, or into a chat where you'll frame it yourself) |
| **Open in ChatGPT / Claude** (menu) | `buildOpenAsk()` | A short, persona-aware ask + the **live URL** (the assistant fetches the page itself) | You want to jump straight into a chat without pasting the whole page |

## The persona system — framing by doc-type

The core idea: the **receiving assistant's role changes with the page's doc-type**, so the copied prompt produces the right behaviour. The preamble is built in three layers:

1. **Persona** — *who the assistant is* ("You are a senior implementation assistant…").
2. **Mission** — *what to do with this page* ("Execute the procedure below in your real workspace…").
3. **Autonomy contract + authentication boundary** — *how far to proceed without asking* (do agent-owned work, update ledgers when state changes, ask only at true human gates; never type passwords, log in, or create accounts).
4. **Rules** — *binding constraints* (steps in order, verify against `# Expected:`, checklist gates, evidence-backed stop points…).

```mermaid
flowchart TD
  A[Doc page] --> B["PageCopyPrompt.astro<br/>builds one meta object"]
  B --> C{"resolvePersona()<br/>path type → kind → shelf"}
  C --> D["Posture<br/>execute · verify · understand · use · evaluate · author"]
  D --> E1["buildPagePrompt()<br/>Copy as prompt"]
  D --> E2["buildOpenAsk()<br/>Open in ChatGPT / Claude"]
  B --> E3["buildPageMarkdown()<br/>Copy page"]
  E1 --> F[CopyPageButton.astro]
  E2 --> F
  E3 --> F
```

### Doc-type → posture

| Shelf · type | Posture | The assistant becomes |
| --- | --- | --- |
| **build** / playbooks · sops · runbooks | `execute` | Senior implementation assistant — runs the steps in your real repo |
| **build** / checklists | `verify` | Verification partner — confirms each item is genuinely done |
| **build** / blueprints | `understand` | Architecture advisor — explains the design before you touch code |
| **learn** / handbooks · guides · concepts · research · case-studies | `understand` | Technical tutor — explains, then helps you apply |
| **resources** / templates · kits · cheatsheets · reference | `use` | Hands-on assistant — adapts the artifact for your project |
| **directory** | `evaluate` | Research assistant — compares options and helps you choose |
| **system_docs** · any `…/meta/authoring/…` page | `author` | ZajLibrary documentation assistant — the **only** place the "documentation assistant" framing is correct |

:::tip[Self-demonstrating]
This page lives under `system_docs/`, so its own **Copy as prompt** gives you the `author` persona — try it.
:::

### Routing precedence

`resolvePersona()` picks the posture in this order (most reliable first):

1. **Authoring override** — `system_docs` shelf or any `…/meta/authoring/…` path → `author`.
2. **Path type segment** — `<shelf>/<type>/…` (the IA enforces this; `content-path-audit.py` validates it).
3. **`kind` frontmatter** — fallback when the path type is unknown (the enum in `content.config.ts`).
4. **Shelf default** — then a neutral `generic` persona.

### Common prompt extras

Every **Copy as prompt** payload includes:

- **Autonomy contract** — the prompt includes `AUTONOMY CONTRACT — complete accurate work unless a real human gate is required`, so a receiving agent does routine agent-owned work itself, updates progress/customization/feedback ledgers when guide state changes, and asks the human only for credentials, production/DNS, destructive operations, legal/commercial/product calls, or genuinely equal options.
- **Authentication boundary** — executable prompts include `AUTHENTICATION BOUNDARY — hard, non-overridable`, so a receiving agent never types passwords, logs in, or creates accounts in any environment. For login-gated checks, the human performs the one login/dashboard sign-in/account creation/credential-entry transaction; the agent resumes after the session exists and drives navigation, screenshots, non-secret fields, and verification.

### Execution extras

For `execute` pages the rules block also gets:

- **Playbook cold-start orientation** — every playbook **Copy as prompt** payload starts with `STEP 0 — COLD-START ORIENTATION`, so a fresh agent first identifies the project, reads ledgers, checks git/concurrent work, inherits project rules, and records feedback before fetching a Part.
- **Part-awareness** — when a document has `partNumber` (or legacy `stepNumber`), a line like *"This is Part 1 of phase 1 (~10 min) — execute this Part fully, then stop at its `## Checklist`."*
- **Category-scoped rules** — CodeCanyon/Laravel pages (`category: laravel` or a `codecanyon` tag/collection) append a **secrets-discipline** rule. Narrow rules stay scoped — they never appear on unrelated pages.
- **Execution coverage contract** — executable Parts with `## Steps` or `## Checklist` append the shared anti-skip contract from `playbook-execution-contract.ts`, requiring every Step, sub-step, checklist box, state update, and server-continuity check to be classified with evidence before an agent can claim done. `MUST`, `SHOULD`, and `OPTIONAL` labels are scope labels, not automatic skip permission; `MUST-only`, `MUST+SHOULD-only`, or fast mode must be an explicit user request and skipped items still need coverage rows.

Runnable playbook Phase overview documents also render `PhaseRunnerPrompt.astro` in two places: a compact **Run this Phase** action beside **Copy as prompt**, and a contextual **Run this Phase** button beside the visible Part cards. Both buttons copy the same Phase prompt with the cold-start orientation first, the autonomy contract next, current-origin URLs plus MCP slugs after it, and the shared execution contract below the ordered Part list. Phase-runner prompts default to **full Phase execution**: every ordered Part URL is in scope, including `SHOULD` / optional-quality Parts, unless the user explicitly asks for `MUST-only`, gate-only, or fast mode. Before fetching Part 1, the copied Phase prompt asks the receiving agent to confirm `FULL` (recommended/default), `MUST-only`, or `MUST+SHOULD-only`, so paste-time behavior is visible.

Numbered playbook Part documents also render `PartRunnerPrompt.astro` as **Run this Part** beside **Copy as prompt**. That button copies a generated one-Part execution prompt, not raw Markdown: cold-start orientation, autonomy/full-coverage clauses, the current-origin URL and MCP slug, terminal-only progress scaffolding, and the shared execution contract. The copied text uses **Part** for the document-level unit and **Step** for each action in `## Steps`.

## How to extend it

| You want to… | Do this in `page-prompt.ts` |
| --- | --- |
| Add a new doc-**type** | Add a row to `TYPE_META` (`{ label, posture }`), and to `KIND_TO_SEGMENT` if a `kind` value should map to it |
| Change a persona's wording | Edit the relevant entry in `POSTURES` (persona / mission / rules / openAsk) |
| Add a category-scoped rule | Extend `CODECANYON_EXECUTE_RULES` (or add a sibling set + a guard in `resolvePersona`) |
| Surface a new metadata field | Add it to `PagePromptMeta`, pass it from `PageCopyPrompt.astro`, and add a `metadataLines` entry |

After any change, run `npm run build` — the persona text ships inside each page's static HTML, so a green build plus a spot-check of the rendered `data-prompt` is the verification.

:::note[Architecture]
`PageCopyPrompt.astro` builds **one** `meta` object and calls all three builders, then renders `CopyPageButton.astro` (which is presentation-only). It mounts in two places: the page title row (`PageTitle.astro`, header placement) and the article footer (`Footer.astro`, footer placement).
:::