microsoft/power-platform-skills / plugins/power-pages/skills/edit-site
edit-site
VerifiedPlans, coordinates and customizes PAC CLI-downloaded Standard or Enhanced data model non-SPA Power Pages sites created from Microsoft templates or manually in Design Studio. Use for requests involving pages, localized content, navigation, assets, snippets, site markers, Liquid/web templates, page layouts, or styling. Coordinated work produces an approved browser-viewable plan, executes changes in dependency order, verifies the local result, and optionally hands off to deployment. Not for React, Angular, Vue, or Astro SPA sites.
references/non-spa-site-authoring.md
View on GitHub# Non-SPA Site Authoring
Shared routing and execution contract for local PAC-downloaded declarative Power Pages
authoring. `/edit-site` supports individual page or component edits and
coordinated customization. These references are not independently invocable skills.
Read-only callers may consult metadata and identity rules without executing authoring.
Work only in the local downloaded site. Do not upload, deploy, activate,
modify Dataverse directly, sync to Studio, or clear caches as part of authoring.
## Dispatch and reference loading
Select the primary workflow from the requested outcome:
| Operation identifier | Requested outcome | Authoritative reference |
|---|---|---|
| `author-web-file` | Import, author, replace, rename, move, configure, or delete a site-served asset and its metadata | [Web files](web-file-authoring.md) |
| `author-content-snippet` | Create, translate, edit, rename, convert, or delete a localized or language-neutral reusable value | [Content snippets](content-snippet-authoring.md) |
| `author-web-template` | Create or modify Liquid/HTML source, fragments, inherited layouts, or header/footer source | [Web templates](web-template-authoring.md) |
| `author-page-template` | Create or modify rendering metadata, an end-to-end custom layout, or associated webpage assignments | [Page templates](page-template-authoring.md) |
| `author-webpage` | Create or modify root/localized records, routes, parents, template assignments, shells, or navigation | [Webpages](webpage-authoring.md) |
| `author-site-marker` | Create, retarget, rename, or locally remove a named same-site webpage binding with verified caller scope | [Site markers](site-marker-authoring.md) |
| `style-site` | Apply guarded CSS/source styling without changing native component structure | [Styling](style-site-workflow.md) |
`operation.skill` and `capabilities[].ownerSkill` use the identifiers in this table to
select reference workflows. Execute them directly in the coordinator.
Do not invoke these operation identifiers through the `Skill` tool.
Do not rewrite an approved plan, its identifiers, or its hash to change dispatch.
New pages update navigation by default unless the user explicitly asks not to.
Resolve placement through the webpage reference before writing. Include navigation placement
or the explicit opt-out in new customization plans; do not expand an approved execution's scope.
Read this index and
`${PLUGIN_ROOT}/references/design-studio-site-discovery.md`, then the complete primary
reference. Load another component reference only when its records or callers are affected.
Do not preload unrelated component references, design guidance, asset planning, or plan contracts for
an individual page or component edit. The component references own PAC fields, serialization, identity, and
verification; do not invent a second interpretation in the entry skill.
Use the body-composition skill and styling workflow:
- `author-webpage-content` composes one exact existing localized body file after its
webpage identity, locale, Bootstrap evidence, values, and preservation are resolved.
- `style-site` owns visual CSS/cascade changes and its separate exact-diff/hash approval.
Execute its reference directly for styling-only requests; do not enter the seven coordinated
phases or invoke it through the `Skill` tool. Asset metadata authoring must not bypass
that workflow.
## Individual page or component edits
Use this route for one resolved component outcome or one straightforward page, including
the exact dependency changes required by its reference. Multiple independently planned
operations, new-site design, redesign, typed output bindings, or resuming an approved
execution use the coordinated workflow instead.
Show **Inspect -> Author -> Independently verify -> Report**. Phase-task bookkeeping,
browser approval documents, and a new Git baseline are not required for these edits.
1. **Locate and inspect.** Follow the shared discovery contract with the primary
reference's required paths. Resolve exactly one declarative site and read the complete
target files. Code-site markers are not sufficient. Preserve pre-existing dirty files.
2. **Classify and resolve.** Establish the action, stable target identity, locale scope,
final values, affected callers, and preservation requirements. Distinguish metadata-only
work from rendering, layout, navigation, and global binding changes.
3. **Resolve dependencies before writing.** Follow the relevant component references.
Stop with candidates and evidence when identity, locale, source, permissions, output
safety, or impact cannot be established. If a reference requires additional approval
that has not been obtained, use the coordinated route and its plan approval before
applying that change; do not infer approval from routine-edit autonomy.
4. **Execute the selected reference workflow.** Generate new IDs through
`${PLUGIN_ROOT}/scripts/generate-uuid.js`; preserve existing IDs and unknown PAC fields.
Make the smallest coherent change. Invoke only the content skill when its execution boundary
is involved, passing the resolved inputs and preview handoff marker. Execute styling through
its reference workflow and exact-diff/hash approval.
5. **Independently verify and report.** Re-read changed files and relationships, run every
applicable reference's verification checklist and prohibited-shortcut checks, and
review the final diff against the operation and locale scope. Report actual files,
stable identities, routes/bindings, dependencies, and any unresolved runtime evidence.
Do not create or replace a customization plan/receipt, initialize Git, commit, or offer
automatic deployment as part of this route. A separately requested deployment
uses `/deploy-site` and its normal identity/environment/approval checks. When a current
approved execution exists, inspect its scope and resume it or establish a fresh coordinated
plan; do not silently edit outside a resumable approval.
## Autonomy and scope
Complete routine authoring without asking the user about routine implementation
details. Infer low-risk file stems, formatting, encoding, accessibility, and creation defaults
from the request, primary reference, site conventions, and comparable records.
Stop before writing when a material choice remains unresolved:
- multiple sites or records with no supported winner;
- missing or ambiguous parent, rendering source, source asset, language, or translation;
- conflicting shared metadata, filename/signature/MIME type, or output context;
- a rename, move, scope/type migration, deletion, or rebind with unresolved dependencies;
- security, visibility, response-boundary, global-binding, or permissions impact that
exceeds the resolved request.
## Coordinated inputs and results
For an approved operation, retain the exact selected site root, action, target identity,
locales, `resolvedInputs`, caller scope, and preservation requirements. Revalidate supplied
paths and existing IDs before writing. Use approved `designContext` and any exact-plan
`imageChecks` without restarting the brief or repeating unchanged network checks.
New-site custom composition does not authorize unrelated changes, record churn, or the
loss of required native behavior. Existing-site narrow edits remain preservation-first.
Execute component operations in the coordinator. Re-read actual results before
recording the stable keys required by
`expectedOutputs`; examples include:
| Workflow | Typical verified results |
|---|---|
| Web file | `publicUrl`, site-relative `metadataPath`, web-file and annotation identities |
| Webpage | `websitePageId`, `route`, exact `localizedTargetFile` for a resolved locale |
| Web/page template | Stable template identity, source/metadata paths, actual bindings |
| Content snippet | Stable record identities, logical name, type, and language scope |
| Site marker | `siteMarkerId`, exact `name`, target `pageId`, and site-relative `metadataPath` |
The approved operation defines the output keys; these examples do not change its contract.
Do not predict generated IDs or URLs, substitute unresolved tokens, or manufacture a missing
output. Complete an operation only after its reference verification succeeds; failure must
remain explicit in the execution receipt.
## Entry-point behavior
The entry skill owns its version check and
`${PLUGIN_ROOT}/references/non-spa-preview.md` disclosure. When invoking a
covered skill, forward `previewNoticeShown: true` and
`invocationSource: edit-site` only after the notice was displayed.
Reading another reference is not another preview entry or skill invocation.
Record successful entry-point usage through
`${PLUGIN_ROOT}/references/skill-tracking-reference.md` as `EditSite`.
Record only skill invocations, not reference reads. Preserve existing tracking records;
never create a tracking directory merely to record non-SPA authoring.
Report any supported tracking diff separately from authored site changes.
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| writing-skillsobra/superpowersUse when creating new skills, editing existing skills, or verifying skills work before deployment | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |
| xlsxanthropics/skillsUse this skill any time a spreadsheet file is the primary input or output. This means any task where the user wants to: open, read, edit, or fix an existing .xlsx, .xlsm, .xltx, .csv, or .tsv file (e.g., adding columns, computing formulas, formatting, charting, cleaning messy data); create a new spreadsheet from scratch or from other data sources; or convert between tabular file formats. Trigger especially when the user references a spreadsheet file by name or path — even casually (like "the xlsx in my downloads") — and wants something done to it or produced from it. Also trigger for cleaning or restructuring messy tabular data files (malformed rows, misplaced headers, junk data) into proper spreadsheets. The deliverable must be a spreadsheet file. Do NOT trigger when the primary deliverable is a Word document, HTML report, standalone Python script, database pipeline, or Google Sheets API integration, even if tabular data is involved. | Stars: 198k | Updated | Claude CodeCodex | Verified | Install |
| shipping-and-launchaddyosmani/agent-skillsPrepares production launches. Use when preparing to deploy to production, or when asking what needs to be in place before shipping. Use when you need a pre-launch checklist, when setting up monitoring, when planning a staged rollout, or when you need a rollback strategy. | Stars: 119k | Updated | Claude CodeCodex | Verified | Install |
| childhoodthedotmack/claude-memBabysit a shipped release after merge, publication, or deployment. Verify published packages, production behavior, installed plugins, and late CI or user reports until the observation window is clean; use babysit for PR readiness before release. | Stars: 118k | Updated | Claude CodeCodex | Verified | Install |