How long does the integration take?

Change one base URL in an OpenAI-compatible client, or install the Python or TypeScript SDK; the quickstart routes a first request in a few minutes.

Will this lock us into another vendor?

No. The router is Apache-2.0 open source, your provider keys stay yours, and the API is OpenAI-compatible, so leaving is a change of base URL.

What happens when a provider fails?

The router falls back to the next model your policy allows and records the fallback in the trace, so an outage costs a retry, not an incident.

visual-updater - Prompt - OpenSmartRoute

Name: visual-updater
Author: aisaac498

Prompt file imported from aisaac498/Jira-Go-Clone (.claude/commands/visual-updater.md). Copyright stays with the author.

/visual-updater

Run a side-by-side visual comparison between the client's live Jira and the clone, then ask the user what needs to be tested or adjusted.

Step 0: Ensure Dev Servers Are Running

Before any visual work, verify both servers are up:

# Check Vite (port 5173)
curl -s -o /dev/null -w "%{http_code}" http://localhost:5173/

# Check Go API (port 8080)
curl -s -o /dev/null -w "%{http_code}" http://localhost:8080/health

If Vite is not running:

cd web && npm run dev &
sleep 4

If the Go API is not running:

export $(grep -v '^#' .env | grep -v '^$' | xargs) && go run ./cmd/api > /tmp/api.log 2>&1 &
sleep 6

If Docker containers are not running:

docker compose up -d
sleep 3

Step 1: Ask the User What to Test

Before opening any browser, ask:

"Which view(s) do you want to compare today?

Options:

Board — Kanban columns, card details, filters, grouping

List — issue table, columns, pagination

Summary — stat cards, charts, activity feed

Issue Detail — fields, subtasks, linked items, activity tabs

Search — global search modal, filters

Create Modal — create issue fields

Sidebar — spaces, recent, starred, navigation

For You — home/dashboard view

All — full walkthrough

Also: is there a specific element or behavior you noticed needs fixing?"

Wait for the user's answer before proceeding. If the user says "all", cover every view in order.

Step 2: Open Both Browsers

Use Playwright MCP to open two tabs:

Tab 1 — Clone (localhost): Navigate to the relevant clone URL based on user's selection:

Board → http://localhost:5173/projects/{id}/board
List → http://localhost:5173/projects/{id}/list
Summary → http://localhost:5173/projects/{id}/summary
Issue Detail → http://localhost:5173/projects/{id}/issues/{issueId}
Search → Open clone, then trigger Cmd+K
Sidebar / For You → http://localhost:5173/dashboard

Tab 2 — Client Jira: Navigate to the relevant client URL:

Board → https://shikhaguleria.atlassian.net/jira/software/projects/NDTL/boards
List → https://shikhaguleria.atlassian.net/jira/software/projects/NDTL/list
Summary → https://shikhaguleria.atlassian.net/jira/software/projects/NDTL/summary
Issue Detail → https://shikhaguleria.atlassian.net/browse/NDTL-116
Search → https://shikhaguleria.atlassian.net/jira/software/projects/NDTL/boards then open search
For You → https://shikhaguleria.atlassian.net/jira/for-you

Note: Atlassian may show popups (Team '26 banner, etc.). Dismiss with Escape key before screenshotting.

Step 3: Take Screenshots

For each view being tested:

Screenshot the clone → save as compare-clone-{view}.png
Screenshot the client Jira → save as compare-jira-{view}.png

Use mcp__plugin_playwright_playwright__browser_take_screenshot for both.

Step 4: Visual Analysis

For each screenshot pair, produce a structured comparison table:

### {View Name} Comparison

| Element | Client Jira | Clone | Gap |
|---------|------------|-------|-----|
| [element] | [what Jira shows] | [what clone shows] | [gap description] |

Rate each gap:

Critical — clone cannot serve the client's workflow without this
High — visible daily, significantly degrades experience
Medium — visible but not blocking
Low — cosmetic or edge case

Step 5: Ask for Adjustment Confirmation

Present the gap list and ask:

"Here are the gaps I found. Which would you like me to fix in this session? I can tackle them in priority order, or you can pick specific items."

Wait for the user's selection before implementing any changes.

Step 6: Implement Selected Fixes

For each confirmed fix:

Implement the change
Take a new screenshot of the clone after the fix
Compare against the Jira screenshot to confirm parity
Report result: ✅ Fixed / ⚠️ Partial / ❌ Blocked (with reason)

Step 7: Summary Report

At the end, output:

## Visual Updater Session Summary

**Tested:** {views tested}
**Fixed:** {N} gaps
**Remaining:** {N} gaps (list with priority)
**Deferred:** {N} gaps (reason: out of scope / Phase 4 / needs backend)

Notes

The client's Jira project key is NDTL (Nuray Dev Task Logs)
Client uses pure Kanban (no sprints, no backlog)
Excluded from comparison: Teams, Dashboards, Apps
Reference doc: docs/superpowers/specs/2026-04-24-client-jira-feature-analysis-design.md
Screenshots from prior audit: jira-live-0*.png at repo root

Install visual-updater and it becomes one more candidate the router can pick - when it fits.

A listing is a routing target with a manifest: what it does, which domains it covers, what it costs and who publishes it. Once installed it sits beside your own models and tools, is scored like any other candidate for each request, and shows up in the trace when it wins. Ratings come from workspaces that installed it, one per account.

# /visual-updater Run a side-by-side visual comparison between the client's live Jira and the clone, then ask the user what needs to be tested or adjusted. ## Step 0: Ensure Dev Servers Are Running Before any visual work, verify both servers are up: ```bash # Check Vite (port 5173) curl -s -o /dev/null -w "%{http_code}" http://localhost:5173/ # Check Go API (port 8080) curl -s -o /dev/null -w "%{http_code}" http://localhost:8080/health ``` If Vite is not running: ```bash cd web && npm run dev & sleep 4 ``` If the Go API is not running: ```bash export $(grep -v '^#' .env | grep -v '^$' | xargs) && go run ./cmd/api > /tmp/api.log 2>&1 & sleep 6 ``` If Docker containers are not running: ```bash docker compose up -d sleep 3 ``` ## Step 1: Ask the User What to Test Before opening any browser, ask: > "Which view(s) do you want to compare today? > > Options: > - **Board** — Kanban columns, card details, filters, grouping > - **List** — issue table, columns, pagination > - **Summary** — stat cards, charts, activity feed > - **Issue Detail** — fields, subtasks, linked items, activity tabs > - **Search** — global search modal, filters > - **Create Modal** — create issue fields > - **Sidebar** — spaces, recent, starred, navigation > - **For You** — home/dashboard view > - **All** — full walkthrough > > Also: is there a specific element or behavior you noticed needs fixing?" Wait for the user's answer before proceeding. If the user says "all", cover every view in order. ## Step 2: Open Both Browsers Use Playwright MCP to open two tabs: **Tab 1 — Clone (localhost):** Navigate to the relevant clone URL based on user's selection: - Board → `http://localhost:5173/projects/{id}/board` - List → `http://localhost:5173/projects/{id}/list` - Summary → `http://localhost:5173/projects/{id}/summary` - Issue Detail → `http://localhost:5173/projects/{id}/issues/{issueId}` - Search → Open clone, then trigger Cmd+K - Sidebar / For You → `http://localhost:5173/dashboard` **Tab 2 — Client Jira:** Navigate to the relevant client URL: - Board → `https://shikhaguleria.atlassian.net/jira/software/projects/NDTL/boards` - List → `https://shikhaguleria.atlassian.net/jira/software/projects/NDTL/list` - Summary → `https://shikhaguleria.atlassian.net/jira/software/projects/NDTL/summary` - Issue Detail → `https://shikhaguleria.atlassian.net/browse/NDTL-116` - Search → `https://shikhaguleria.atlassian.net/jira/software/projects/NDTL/boards` then open search - For You → `https://shikhaguleria.atlassian.net/jira/for-you` > Note: Atlassian may show popups (Team '26 banner, etc.). Dismiss with Escape key before screenshotting. ## Step 3: Take Screenshots For each view being tested: 1. Screenshot the **clone** → save as `compare-clone-{view}.png` 2. Screenshot the **client Jira** → save as `compare-jira-{view}.png` Use `mcp__plugin_playwright_playwright__browser_take_screenshot` for both. ## Step 4: Visual Analysis For each screenshot pair, produce a structured comparison table: ```markdown ### {View Name} Comparison | Element | Client Jira | Clone | Gap | |---------|------------|-------|-----| | [element] | [what Jira shows] | [what clone shows] | [gap description] | ``` Rate each gap: - **Critical** — clone cannot serve the client's workflow without this - **High** — visible daily, significantly degrades experience - **Medium** — visible but not blocking - **Low** — cosmetic or edge case ## Step 5: Ask for Adjustment Confirmation Present the gap list and ask: > "Here are the gaps I found. Which would you like me to fix in this session? > I can tackle them in priority order, or you can pick specific items." Wait for the user's selection before implementing any changes. ## Step 6: Implement Selected Fixes For each confirmed fix: 1. Implement the change 2. Take a new screenshot of the clone after the fix 3. Compare against the Jira screenshot to confirm parity 4. Report result: ✅ Fixed / ⚠️ Partial / ❌ Blocked (with reason) ## Step 7: Summary Report At the end, output: ```markdown ## Visual Updater Session Summary **Tested:** {views tested} **Fixed:** {N} gaps **Remaining:** {N} gaps (list with priority) **Deferred:** {N} gaps (reason: out of scope / Phase 4 / needs backend) ``` --- ## Notes - The client's Jira project key is **NDTL** (Nuray Dev Task Logs) - Client uses pure Kanban (no sprints, no backlog) - Excluded from comparison: Teams, Dashboards, Apps - Reference doc: `docs/superpowers/specs/2026-04-24-client-jira-feature-analysis-design.md` - Screenshots from prior audit: `jira-live-0*.png` at repo root

{ "prompt": "# /visual-updater\n\nRun a side-by-side visual comparison between the client's live Jira and the clone, then ask the user what needs to be tested or adjusted.\n\n## Step 0: Ensure Dev Servers Are Running\n\nBefore any visual work, verify both servers are up:\n\n```bash\n# Check Vite (port 5173)\ncurl -s -o /dev/null -w \"%{http_code}\" http://localhost:5173/\n\n# Check Go API (port 8080)\ncurl -s -o /dev/null -w \"%{http_code}\" http://localhost:8080/health\n```\n\nIf Vite is not running:\n```bash\ncd web && npm run dev &\nsleep 4\n```\n\nIf the Go API is not running:\n```bash\nexport $(grep -v '^#' .env | grep -v '^$' | xargs) && go run ./cmd/api > /tmp/api.log 2>&1 &\nsleep 6\n```\n\nIf Docker containers are not running:\n```bash\ndocker compose up -d\nsleep 3\n```\n\n## Step 1: Ask the User What to Test\n\nBefore opening any browser, ask:\n\n> \"Which view(s) do you want to compare today?\n>\n> Options:\n> - **Board** — Kanban columns, card details, filters, grouping\n> - **List** — issue table, columns, pagination\n> - **Summary** — stat cards, charts, activity feed\n> - **Issue Detail** — fields, subtasks, linked items, activity tabs\n> - **Search** — global search modal, filters\n> - **Create Modal** — create issue fields\n> - **Sidebar** — spaces, recent, starred, navigation\n> - **For You** — home/dashboard view\n> - **All** — full walkthrough\n>\n> Also: is there a specific element or behavior you noticed needs fixing?\"\n\nWait for the user's answer before proceeding. If the user says \"all\", cover every view in order.\n\n## Step 2: Open Both Browsers\n\nUse Playwright MCP to open two tabs:\n\n**Tab 1 — Clone (localhost):**\nNavigate to the relevant clone URL based on user's selection:\n- Board → `http://localhost:5173/projects/{id}/board`\n- List → `http://localhost:5173/projects/{id}/list`\n- Summary → `http://localhost:5173/projects/{id}/summary`\n- Issue Detail → `http://localhost:5173/projects/{id}/issues/{issueId}`\n- Search → Open clone, then trigger Cmd+K\n- Sidebar / For You → `http://localhost:5173/dashboard`\n\n**Tab 2 — Client Jira:**\nNavigate to the relevant client URL:\n- Board → `https://shikhaguleria.atlassian.net/jira/software/projects/NDTL/boards`\n- List → `https://shikhaguleria.atlassian.net/jira/software/projects/NDTL/list`\n- Summary → `https://shikhaguleria.atlassian.net/jira/software/projects/NDTL/summary`\n- Issue Detail → `https://shikhaguleria.atlassian.net/browse/NDTL-116`\n- Search → `https://shikhaguleria.atlassian.net/jira/software/projects/NDTL/boards` then open search\n- For You → `https://shikhaguleria.atlassian.net/jira/for-you`\n\n> Note: Atlassian may show popups (Team '26 banner, etc.). Dismiss with Escape key before screenshotting.\n\n## Step 3: Take Screenshots\n\nFor each view being tested:\n\n1. Screenshot the **clone** → save as `compare-clone-{view}.png`\n2. Screenshot the **client Jira** → save as `compare-jira-{view}.png`\n\nUse `mcp__plugin_playwright_playwright__browser_take_screenshot` for both.\n\n## Step 4: Visual Analysis\n\nFor each screenshot pair, produce a structured comparison table:\n\n```markdown\n### {View Name} Comparison\n\n| Element | Client Jira | Clone | Gap |\n|---------|------------|-------|-----|\n| [element] | [what Jira shows] | [what clone shows] | [gap description] |\n```\n\nRate each gap:\n- **Critical** — clone cannot serve the client's workflow without this\n- **High** — visible daily, significantly degrades experience\n- **Medium** — visible but not blocking\n- **Low** — cosmetic or edge case\n\n## Step 5: Ask for Adjustment Confirmation\n\nPresent the gap list and ask:\n\n> \"Here are the gaps I found. Which would you like me to fix in this session?\n> I can tackle them in priority order, or you can pick specific items.\"\n\nWait for the user's selection before implementing any changes.\n\n## Step 6: Implement Selected Fixes\n\nFor each confirmed fix:\n1. Implement the change\n2. Take a new screenshot of the clone after the fix\n3. Compare against the Jira screenshot to confirm parity\n4. Report result: ✅ Fixed / ⚠️ Partial / ❌ Blocked (with reason)\n\n## Step 7: Summary Report\n\nAt the end, output:\n\n```markdown\n## Visual Updater Session Summary\n\n**Tested:** {views tested}\n**Fixed:** {N} gaps\n**Remaining:** {N} gaps (list with priority)\n**Deferred:** {N} gaps (reason: out of scope / Phase 4 / needs backend)\n```\n\n---\n\n## Notes\n\n- The client's Jira project key is **NDTL** (Nuray Dev Task Logs)\n- Client uses pure Kanban (no sprints, no backlog)\n- Excluded from comparison: Teams, Dashboards, Apps\n- Reference doc: `docs/superpowers/specs/2026-04-24-client-jira-feature-analysis-design.md`\n- Screenshots from prior audit: `jira-live-0*.png` at repo root", "variables": [], "notes": "Run a side-by-side visual comparison between the client's live Jira and the clone, then ask the user what needs to be tested or adjusted.", "metadata": { "source": { "provider": "github-claude-commands", "repository": "https://github.com/aisaac498/Jira-Go-Clone", "path": ".claude/commands/visual-updater.md", "ref": "e6484863ac712f674826626b695c54d79e8913c1", "url": "https://github.com/aisaac498/Jira-Go-Clone/blob/e6484863ac712f674826626b695c54d79e8913c1/.claude/commands/visual-updater.md", "key": "aisaac498/Jira-Go-Clone/.claude/commands/visual-updater.md" } } }

visual-updater

About

/visual-updater

Step 0: Ensure Dev Servers Are Running

Step 1: Ask the User What to Test

Step 2: Open Both Browsers

Step 3: Take Screenshots

Step 4: Visual Analysis

Step 5: Ask for Adjustment Confirmation

Step 6: Implement Selected Fixes

Step 7: Summary Report

Notes

Install visual-updater and it becomes one more candidate the router can pick - when it fits.

What it costs to adopt - in engineering time.

One base URL

No vendor lock-in

Failover built in

Use it

Manifest

Reviews