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.

check - Prompt - OpenSmartRoute

Name: check
Author: oharms

Prompt file imported from oharms/PanWizard (.claude/commands/check.md). Copyright stays with the author.

/check - Verify PAN Wizard Implementations

Verify that PAN Wizard features actually work, not just exist.

⛔ Self-Protection Gate

This is the PAN Wizard SOURCE REPOSITORY. NEVER install PAN here. Installation verification goes to d:\pantesting.

Usage

/check                       # Verify current feature (auto-detect from context)
/check --verify <feature>    # Deep implementation verification
/check --install             # Verify installer works end-to-end
/check --all                 # Full verification (tests + install)

--verify: Implementation Verification

Steps

Identify what to validate (from argument or recent changes)
Find implementation files (installer, core libs, commands, agents)
Read the code to understand what it claims to do
Run unit tests that cover this feature:
```
npm test
```
Run scenario tests — they cover the installer and also the pan-tools contracts and workflows, so run them for any engine change too:
```
npm run test:scenarios
```
Run learn-lint to catch learnings-store integrity drift (duplicate IDs, dangling refs, scope leaks):
```
node pan-wizard-core/bin/pan-tools.cjs learn lint --raw
```
Exits non-zero on errors; warnings are advisory.

Manual verification if needed — install to test directory:

cd d:\pantesting && node d:\PanWizard\bin\install.js --all --local

Report verified vs. failing features with evidence

Severity Levels

Level	Meaning	Action
VERIFIED ✅	Feature works as claimed	None
PARTIAL ⚠️	Works for some runtimes or partial functionality	Fix gaps
BROKEN ❌	Claimed but doesn't work	Fix or mark as in-progress
MISSING 🔴	Not implemented at all	Remove false claim

--install: Installer Verification

Verify all 5 runtimes install correctly:

# Clean slate
Remove-Item d:\pantesting\.claude, d:\pantesting\.codex, d:\pantesting\.gemini, d:\pantesting\.opencode, d:\pantesting\.github, d:\pantesting\.agents, d:\pantesting\AGENTS.md, d:\pantesting\.mcp.json -Recurse -Force -ErrorAction SilentlyContinue

# Install each runtime
cd d:\pantesting
node d:\PanWizard\bin\install.js --claude --local
node d:\PanWizard\bin\install.js --codex --local
node d:\PanWizard\bin\install.js --gemini --local
node d:\PanWizard\bin\install.js --opencode --local
node d:\PanWizard\bin\install.js --copilot --local

Verify each installed correctly:

Commands present: commands/pan/ for Claude (.md) and Gemini (.toml), flat commands/pan-*.md for OpenCode, skills/pan-*/ for Copilot, and .agents/skills/pan-*/ at the project root for Codex
Agents present in agents/ (Codex writes .toml, Copilot .agent.md)
Core modules present (pan-wizard-core/bin/)
Hooks present in hooks/ for every runtime except OpenCode, which gets plugins/pan-wizard.js instead
Manifest present (pan-file-manifest.json)

--all: Combined Check

Runs --verify (tests) then --install (end-to-end).

Never Do

Report "VERIFIED" without running tests or checking actual behavior
Skip installer verification (all 5 runtimes must be checked)
Change test expectations to match broken behavior
Trust documentation claims without execution evidence

Install check 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.

# /check - Verify PAN Wizard Implementations Verify that PAN Wizard features actually work, not just exist. ## ⛔ Self-Protection Gate This is the PAN Wizard SOURCE REPOSITORY. NEVER install PAN here. Installation verification goes to `d:\pantesting`. --- ## Usage ``` /check # Verify current feature (auto-detect from context) /check --verify <feature> # Deep implementation verification /check --install # Verify installer works end-to-end /check --all # Full verification (tests + install) ``` --- ## --verify: Implementation Verification ### Steps 1. **Identify** what to validate (from argument or recent changes) 2. **Find** implementation files (installer, core libs, commands, agents) 3. **Read the code** to understand what it claims to do 4. **Run unit tests** that cover this feature: ```powershell npm test ``` 5. **Run scenario tests** — they cover the installer and also the `pan-tools` contracts and workflows, so run them for any engine change too: ```powershell npm run test:scenarios ``` 6. **Run learn-lint** to catch learnings-store integrity drift (duplicate IDs, dangling refs, scope leaks): ```powershell node pan-wizard-core/bin/pan-tools.cjs learn lint --raw ``` Exits non-zero on errors; warnings are advisory. 7. **Manual verification** if needed — install to test directory: ```powershell cd d:\pantesting && node d:\PanWizard\bin\install.js --all --local ``` 8. **Report** verified vs. failing features with evidence ### Severity Levels | Level | Meaning | Action | |-------|---------|--------| | VERIFIED ✅ | Feature works as claimed | None | | PARTIAL ⚠️ | Works for some runtimes or partial functionality | Fix gaps | | BROKEN ❌ | Claimed but doesn't work | Fix or mark as in-progress | | MISSING 🔴 | Not implemented at all | Remove false claim | --- ## --install: Installer Verification Verify all 5 runtimes install correctly: ```powershell # Clean slate Remove-Item d:\pantesting\.claude, d:\pantesting\.codex, d:\pantesting\.gemini, d:\pantesting\.opencode, d:\pantesting\.github, d:\pantesting\.agents, d:\pantesting\AGENTS.md, d:\pantesting\.mcp.json -Recurse -Force -ErrorAction SilentlyContinue # Install each runtime cd d:\pantesting node d:\PanWizard\bin\install.js --claude --local node d:\PanWizard\bin\install.js --codex --local node d:\PanWizard\bin\install.js --gemini --local node d:\PanWizard\bin\install.js --opencode --local node d:\PanWizard\bin\install.js --copilot --local ``` Verify each installed correctly: - Commands present: `commands/pan/` for Claude (`.md`) and Gemini (`.toml`), flat `commands/pan-*.md` for OpenCode, `skills/pan-*/` for Copilot, and `.agents/skills/pan-*/` at the project root for Codex - Agents present in `agents/` (Codex writes `.toml`, Copilot `.agent.md`) - Core modules present (`pan-wizard-core/bin/`) - Hooks present in `hooks/` for every runtime except OpenCode, which gets `plugins/pan-wizard.js` instead - Manifest present (`pan-file-manifest.json`) --- ## --all: Combined Check Runs `--verify` (tests) then `--install` (end-to-end). --- ## Never Do - Report "VERIFIED" without running tests or checking actual behavior - Skip installer verification (all 5 runtimes must be checked) - Change test expectations to match broken behavior - Trust documentation claims without execution evidence

{ "prompt": "# /check - Verify PAN Wizard Implementations\n\nVerify that PAN Wizard features actually work, not just exist.\n\n## ⛔ Self-Protection Gate\n\nThis is the PAN Wizard SOURCE REPOSITORY. NEVER install PAN here.\nInstallation verification goes to `d:\\pantesting`.\n\n---\n\n## Usage\n\n```\n/check # Verify current feature (auto-detect from context)\n/check --verify <feature> # Deep implementation verification\n/check --install # Verify installer works end-to-end\n/check --all # Full verification (tests + install)\n```\n\n---\n\n## --verify: Implementation Verification\n\n### Steps\n\n1. **Identify** what to validate (from argument or recent changes)\n2. **Find** implementation files (installer, core libs, commands, agents)\n3. **Read the code** to understand what it claims to do\n4. **Run unit tests** that cover this feature:\n ```powershell\n npm test\n ```\n5. **Run scenario tests** — they cover the installer and also the `pan-tools` contracts and workflows, so run them for any engine change too:\n ```powershell\n npm run test:scenarios\n ```\n6. **Run learn-lint** to catch learnings-store integrity drift (duplicate IDs, dangling refs, scope leaks):\n ```powershell\n node pan-wizard-core/bin/pan-tools.cjs learn lint --raw\n ```\n Exits non-zero on errors; warnings are advisory.\n7. **Manual verification** if needed — install to test directory:\n ```powershell\n cd d:\\pantesting && node d:\\PanWizard\\bin\\install.js --all --local\n ```\n8. **Report** verified vs. failing features with evidence\n\n### Severity Levels\n\n| Level | Meaning | Action |\n|-------|---------|--------|\n| VERIFIED ✅ | Feature works as claimed | None |\n| PARTIAL ⚠️ | Works for some runtimes or partial functionality | Fix gaps |\n| BROKEN ❌ | Claimed but doesn't work | Fix or mark as in-progress |\n| MISSING 🔴 | Not implemented at all | Remove false claim |\n\n---\n\n## --install: Installer Verification\n\nVerify all 5 runtimes install correctly:\n\n```powershell\n# Clean slate\nRemove-Item d:\\pantesting\\.claude, d:\\pantesting\\.codex, d:\\pantesting\\.gemini, d:\\pantesting\\.opencode, d:\\pantesting\\.github, d:\\pantesting\\.agents, d:\\pantesting\\AGENTS.md, d:\\pantesting\\.mcp.json -Recurse -Force -ErrorAction SilentlyContinue\n\n# Install each runtime\ncd d:\\pantesting\nnode d:\\PanWizard\\bin\\install.js --claude --local\nnode d:\\PanWizard\\bin\\install.js --codex --local\nnode d:\\PanWizard\\bin\\install.js --gemini --local\nnode d:\\PanWizard\\bin\\install.js --opencode --local\nnode d:\\PanWizard\\bin\\install.js --copilot --local\n```\n\nVerify each installed correctly:\n- Commands present: `commands/pan/` for Claude (`.md`) and Gemini (`.toml`), flat `commands/pan-*.md` for OpenCode, `skills/pan-*/` for Copilot, and `.agents/skills/pan-*/` at the project root for Codex\n- Agents present in `agents/` (Codex writes `.toml`, Copilot `.agent.md`)\n- Core modules present (`pan-wizard-core/bin/`)\n- Hooks present in `hooks/` for every runtime except OpenCode, which gets `plugins/pan-wizard.js` instead\n- Manifest present (`pan-file-manifest.json`)\n\n---\n\n## --all: Combined Check\n\nRuns `--verify` (tests) then `--install` (end-to-end).\n\n---\n\n## Never Do\n\n- Report \"VERIFIED\" without running tests or checking actual behavior\n- Skip installer verification (all 5 runtimes must be checked)\n- Change test expectations to match broken behavior\n- Trust documentation claims without execution evidence", "variables": [], "notes": "Verify that PAN Wizard features actually work, not just exist.", "metadata": { "source": { "provider": "github-claude-commands", "repository": "https://github.com/oharms/PanWizard", "path": ".claude/commands/check.md", "ref": "8554d3f8af029bba0c38ab19576b66f76ee71595", "url": "https://github.com/oharms/PanWizard/blob/8554d3f8af029bba0c38ab19576b66f76ee71595/.claude/commands/check.md", "key": "oharms/PanWizard/.claude/commands/check.md" } } }

check

About

/check - Verify PAN Wizard Implementations

⛔ Self-Protection Gate

Usage

--verify: Implementation Verification

Steps

Severity Levels

--install: Installer Verification

--all: Combined Check

Never Do

Install check 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