sorryhyun/apps/self-inspection

Self Inspection (Verb Mode)

A deep diagnostic suite that stress-tests YAAR's verb layer — identical coverage to Self Inspection but using only the 5 generic verbs (describe, read, list, invoke, delete) against yaar:// URIs.

Launch

This is a pure-skill app — no iframe or compiled code. Follow the instructions below using verb tools directly.

When the user opens this app or says "run verb inspection", "run verb diagnostics", or "verb self-test", run all checks below and produce the report.

Important: Verb mode must be enabled (verbMode: true in settings). If it's not, enable it first:

invoke('yaar://config/settings', { verbMode: true })

---

Diagnostics

1. Describe & Discovery

Test that describe returns valid schemas for key resources across multiple domains:

describe('yaar://config/settings')
describe('yaar://storage')
describe('yaar://session/monitors')
describe('yaar://user/notifications')
describe('yaar://apps')
describe('yaar://session/agents')
describe('yaar://session')

PASS if all 7 describe calls return verbs arrays and descriptions without errors.

2. Session Root & Namespace Listing

Test the root resource and namespace enumeration:

read('yaar://')                    # should return session overview (sessionId, platform, etc.)
list('yaar://')                    # should return all 6 URI namespaces (apps, storage, windows, config, browser, sessions)
list('yaar://config/')             # should return config sections (settings, hooks, shortcuts, mounts, app)

PASS if root read returns session info and both list calls return expected items.

3. Concurrent Multi-Renderer Windows

Open 5 windows simultaneously using different renderers via invoke, verify all exist via list, then close all via delete:

invoke('yaar://windows/si-v-md', { action: "create", title: "Test: Markdown", width: 300, height: 200, renderer: "markdown", content: "# Markdown\n\n**Bold** and *italic*." })
invoke('yaar://windows/si-v-html', { action: "create", title: "Test: HTML", width: 300, height: 200, renderer: "html", content: "<div style='padding:16px'><h2>HTML</h2><p style='color:green'>Styled content.</p></div>" })
invoke('yaar://windows/si-v-text', { action: "create", title: "Test: Text", width: 300, height: 200, renderer: "text", content: "Plain text content.\nLine 2.\nLine 3." })
invoke('yaar://windows/si-v-comp', { action: "create", title: "Test: Component", width: 300, height: 200, renderer: "component", content: { components: [{ "type": "text", "content": "Component DSL", "variant": "heading" }, { "type": "badge", "label": "OK", "variant": "success" }, { "type": "progress", "value": 75, "label": "Progress" }] } })
invoke('yaar://windows/si-v-tbl', { action: "create", title: "Test: Table", width: 300, height: 200, renderer: "table", content: { "headers": ["Col A", "Col B"], "rows": [["1", "2"], ["3", "4"]] } })

Verify:

list('yaar://session/monitors')   # all 5 windows should appear: si-v-md, si-v-html, si-v-text, si-v-comp, si-v-tbl

PASS if all 5 windows appear in the list. Then close all:

delete('yaar://windows/si-v-md')
delete('yaar://windows/si-v-html')
delete('yaar://windows/si-v-text')
delete('yaar://windows/si-v-comp')
delete('yaar://windows/si-v-tbl')

4. Window Content Update Operations

Create a markdown window and test all update operations via invoke:

invoke('yaar://windows/si-v-update', { action: "create", title: "Update Test", renderer: "markdown", content: "Line 1" })
invoke('yaar://windows/si-v-update', { action: "update", operation: "append", content: "\nLine 2" })
read('yaar://windows/si-v-update')          # should contain "Line 1\nLine 2"
invoke('yaar://windows/si-v-update', { action: "update", operation: "prepend", content: "Line 0\n" })
read('yaar://windows/si-v-update')          # should contain "Line 0\nLine 1\nLine 2"
invoke('yaar://windows/si-v-update', { action: "update", operation: "replace", content: "Replaced." })
read('yaar://windows/si-v-update')          # should contain only "Replaced."
delete('yaar://windows/si-v-update')

PASS if each read returns the expected content after each operation.

5. Window Lock/Unlock

Lock semantics: a locked window can only be modified by the agent that locked it. Other agents are rejected.

invoke('yaar://windows/si-v-lock', { action: "create", title: "Lock Test", renderer: "text", content: "Locked window test." })
invoke('yaar://windows/si-v-lock', { action: "lock" })
read('yaar://windows/si-v-lock')            # verify locked: true and lockedBy matches current agent

The locking agent can still update its own locked window:

invoke('yaar://windows/si-v-lock', { action: "update", operation: "replace", content: "Updated by locking agent." })
read('yaar://windows/si-v-lock')            # should contain "Updated by locking agent."

Unlock and verify state change:

invoke('yaar://windows/si-v-lock', { action: "unlock" })
read('yaar://windows/si-v-lock')            # verify locked: false
delete('yaar://windows/si-v-lock')

PASS if lock metadata is correct, locking agent can update, and unlock clears the lock.

6. Component Form Submission (interactive)

Create a component window with a form and ask the user to fill it:

invoke('yaar://windows/si-v-form', {
  action: "create",
  title: "Form Test",
  renderer: "component",
  width: 400, height: 300,
  content: {
    components: [
      { "type": "text", "content": "Fill out this form and click Submit.", "variant": "heading" },
      { "type": "input", "name": "username", "formId": "test-form", "label": "Username", "placeholder": "Enter anything" },
      { "type": "select", "name": "color", "formId": "test-form", "label": "Favorite Color", "options": [{ "value": "red", "label": "Red" }, { "value": "blue", "label": "Blue" }, { "value": "green", "label": "Green" }] },
      { "type": "input", "name": "notes", "formId": "test-form", "label": "Notes", "placeholder": "Optional", "rows": 2 },
      { "type": "button", "label": "Submit", "submitForm": "test-form", "action": "form-submitted", "variant": "primary" }
    ]
  }
})

Tell the user: "Please fill out the form and click Submit." When you receive the COMPONENT_ACTION event with form data, read the submitted values and close the window. PASS if form data is received with username and color fields.

delete('yaar://windows/si-v-form')

7. App Protocol Round-Trip (Excel)

Open Excel Lite, query its manifest, write cells, read them back, verify data integrity:

invoke('yaar://windows/si-v-excel', { action: "create", title: "Excel Lite", appId: "excel-lite", renderer: "iframe", content: "yaar://apps/excel-lite" })

Wait for App Protocol ready, then query manifest:

invoke('yaar://windows/si-v-excel', { action: "app_query" })

Verify manifest contains setCells command and cells state key.

Write test data:

invoke('yaar://windows/si-v-excel', { action: "app_command", command: "setCells", params: { "cells": { "A1": "Name", "B1": "Score", "A2": "Alice", "B2": "95", "A3": "Bob", "B3": "87" } } })

Read back:

invoke('yaar://windows/si-v-excel', { action: "app_query", stateKey: "cells" })

Verify cells A1="Name", B1="Score", A2="Alice", B2="95", A3="Bob", B3="87".

Test clearRange:

invoke('yaar://windows/si-v-excel', { action: "app_command", command: "clearRange", params: { "start": "A3", "end": "B3" } })
invoke('yaar://windows/si-v-excel', { action: "app_query", stateKey: "cells" })

Verify A3 and B3 are now empty/missing, but A1-B2 still intact.

delete('yaar://windows/si-v-excel')

PASS if all read-back values match expectations.

8. App Protocol Round-Trip (Word)

Open Word Lite, set content, read it back:

invoke('yaar://windows/si-v-word', { action: "create", title: "Word Lite", appId: "word-lite", renderer: "iframe", content: "yaar://apps/word-lite" })

Wait for ready, then:

invoke('yaar://windows/si-v-word', { action: "app_command", command: "setTitle", params: { "title": "Self Inspection Test" } })
invoke('yaar://windows/si-v-word', { action: "app_command", command: "setContent", params: { "content": "<h1>Test Document</h1><p>This is a self-inspection test.</p>", "renderer": "html" } })
invoke('yaar://windows/si-v-word', { action: "app_query", stateKey: "title" })   # should be "Self Inspection Test"
invoke('yaar://windows/si-v-word', { action: "app_query", stateKey: "stats" })    # should have words > 0
invoke('yaar://windows/si-v-word', { action: "app_query", stateKey: "text" })     # should contain "self-inspection test"
delete('yaar://windows/si-v-word')

PASS if title, stats, and text match expectations.

9. Cross-App Data Flow (Storage → Excel)

Write structured data to storage via invoke, then import it into Excel via App Protocol:

invoke('yaar://storage/_si-v-test-data.json', { action: "write", content: "{\"cells\":{\"A1\":\"Product\",\"B1\":\"Price\",\"A2\":\"Widget\",\"B2\":\"9.99\",\"A3\":\"Gadget\",\"B3\":\"24.50\"},\"styles\":{\"A1\":{\"bold\":true},\"B1\":{\"bold\":true}}}" })
read('yaar://storage/_si-v-test-data.json')     # verify JSON is readable

Open Excel and import:

invoke('yaar://windows/si-v-cross', { action: "create", title: "Cross-App Test", appId: "excel-lite", renderer: "iframe", content: "yaar://apps/excel-lite" })

Wait for ready, then import the data you read from storage:

invoke('yaar://windows/si-v-cross', { action: "app_command", command: "importWorkbook", params: { "data": <parsed JSON from storage read> } })
invoke('yaar://windows/si-v-cross', { action: "app_query", stateKey: "cells" })

Verify A1="Product", B2="9.99".

Cleanup:

delete('yaar://windows/si-v-cross')
delete('yaar://storage/_si-v-test-data.json')

PASS if imported data matches the original JSON.

10. Shortcut Create/Delete via Config URI

invoke('yaar://config/shortcuts', { label: "SI Verb Test", icon: "🧪", shortcutType: "action", target: "self-inspection test" })
read('yaar://config/shortcuts')     # verify shortcut appears, note the shortcutId
delete('yaar://config/shortcuts/<the-shortcut-id>')
read('yaar://config/shortcuts')     # verify it's gone

PASS if shortcut appears after create and disappears after remove.

11. Component Update

Create a component window, then replace its entire layout:

invoke('yaar://windows/si-v-comp-upd', {
  action: "create",
  title: "Component Update Test",
  renderer: "component",
  width: 350, height: 200,
  content: {
    components: [
      { "type": "text", "content": "Version 1", "variant": "heading" },
      { "type": "progress", "value": 25, "label": "Progress" }
    ]
  }
})

Update the components:

invoke('yaar://windows/si-v-comp-upd', {
  action: "update",
  operation: "replace",
  renderer: "component",
  content: {
    components: [
      { "type": "text", "content": "Version 2", "variant": "heading" },
      { "type": "progress", "value": 100, "label": "Complete", "variant": "success" },
      { "type": "badge", "label": "Updated", "variant": "info" }
    ]
  }
})

delete('yaar://windows/si-v-comp-upd')

PASS if component update succeeds without error.

12. Storage Directory Operations via URI

Test directory creation and listing:

invoke('yaar://storage/_si-v-test-dir/file1.txt', { action: "write", content: "File 1" })
invoke('yaar://storage/_si-v-test-dir/file2.txt', { action: "write", content: "File 2" })
invoke('yaar://storage/_si-v-test-dir/sub/file3.txt', { action: "write", content: "File 3" })
list('yaar://storage/_si-v-test-dir')           # should show file1.txt, file2.txt, sub/
list('yaar://storage/_si-v-test-dir/sub')       # should show file3.txt
delete('yaar://storage/_si-v-test-dir/file1.txt')
delete('yaar://storage/_si-v-test-dir/file2.txt')
delete('yaar://storage/_si-v-test-dir/sub/file3.txt')

PASS if directory listing shows expected files and subdirectory.

13. Multi-App Simultaneous

Open 3 App Protocol apps simultaneously and interact with all of them:

invoke('yaar://windows/si-v-multi-excel', { action: "create", title: "Multi: Excel", appId: "excel-lite", renderer: "iframe", content: "yaar://apps/excel-lite" })
invoke('yaar://windows/si-v-multi-word', { action: "create", title: "Multi: Word", appId: "word-lite", renderer: "iframe", content: "yaar://apps/word-lite" })
invoke('yaar://windows/si-v-multi-img', { action: "create", title: "Multi: Images", appId: "image-viewer", renderer: "iframe", content: "yaar://apps/image-viewer" })

Wait for all 3 to be ready, then interact with each:

invoke('yaar://windows/si-v-multi-excel', { action: "app_command", command: "setCells", params: { "cells": { "A1": "Multi-app test" } } })
invoke('yaar://windows/si-v-multi-word', { action: "app_command", command: "setContent", params: { "content": "<p>Multi-app test</p>", "renderer": "html" } })
invoke('yaar://windows/si-v-multi-img', { action: "app_command", command: "setLayout", params: { "mode": "grid", "columns": 3 } })

Query each to verify:

invoke('yaar://windows/si-v-multi-excel', { action: "app_query", stateKey: "cells" })    # A1 = "Multi-app test"
invoke('yaar://windows/si-v-multi-word', { action: "app_query", stateKey: "text" })      # contains "Multi-app test"
invoke('yaar://windows/si-v-multi-img', { action: "app_query", stateKey: "layout" })     # mode = "grid"

Close all:

delete('yaar://windows/si-v-multi-excel')
delete('yaar://windows/si-v-multi-word')
delete('yaar://windows/si-v-multi-img')

PASS if all 3 apps respond correctly to commands and queries simultaneously.

14. Monitor-as-Resource

Read the monitor resource to verify it returns status:

read('yaar://session/monitors')

PASS if returns monitorId, hasMainAgent, windows list, and stats.

15. Agents Discovery

List and inspect active agents:

list('yaar://session/agents')

If agents are listed, read one:

read('yaar://session/agents/<first-agent-id>')

PASS if agent list returns without error and (if agents exist) individual agent read returns agent info.

16. User Notifications via URI

Test notification lifecycle through verb layer:

invoke('yaar://user/notifications', { title: "Verb Test", body: "Self-inspection notification test", variant: "info" })

PASS if notification is shown without error.

---

Report Format

After all checks, create a markdown window with the results:

invoke('yaar://windows/self-inspection-report', {
  action: "create",
  title: "Self Inspection Report (Verb Mode)",
  width: 750, height: 750,
  renderer: "markdown",
  content: "# Self Inspection Report (Verb Mode)\n\n| # | Check | Status | Details |\n|---|-------|--------|---------|\n| 1 | Describe & Discovery | PASS | 7/7 resources described |\n| 2 | Session Root & Namespaces | PASS | root read + 2 list calls verified |\n| 3 | Multi-Renderer Windows | PASS | 5/5 renderers created |\n| 4 | Content Updates | PASS | append/prepend/replace verified |\n| 5 | Window Lock/Unlock | PASS | lock metadata correct, owner update allowed, unlock cleared |\n| 6 | Form Submission | PASS | received username, color fields |\n| 7 | App Protocol (Excel) | PASS | setCells/query/clearRange verified |\n| 8 | App Protocol (Word) | PASS | setContent/title/stats verified |\n| 9 | Cross-App Data Flow | PASS | storage → excel import verified |\n| 10 | Shortcuts via Config URI | PASS | create/list/remove verified |\n| 11 | Component Update | PASS | layout replaced successfully |\n| 12 | Storage Directories | PASS | nested dirs and listing verified |\n| 13 | Multi-App Simultaneous | PASS | 3 apps commanded simultaneously |\n| 14 | Monitor-as-Resource | PASS | monitor status returned |\n| 15 | Agents Discovery | PASS | agent list/read verified |\n| 16 | User Notifications | PASS | notification shown |\n\n**Result: X/16 checks passed**\n\n### Verb Coverage\n| Verb | Tested In |\n|------|-----------|\n| describe | #1 |\n| read | #2, #4, #5, #14, #15 |\n| list | #2, #3, #12, #15 |\n| invoke | #3–#13, #16 |\n| delete | #3–#5, #9, #10, #12, #13 |"
})

Mark each check as:

• PASS — expected result received
• FAIL — unexpected result or error (include error message in Details)
• SKIP — could not run (explain why in Details)

Cleanup Guarantee

If any test fails partway through, always attempt cleanup (close windows, delete test files, remove test apps). Never leave test artifacts behind. Window IDs used by this inspection all start with si-v- for easy identification.

Verb Coverage Notes

All diagnostics use the 5 generic verbs exclusively (describe, read, list, invoke, delete) against yaar:// URIs.