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.

test-runner - Agent - OpenSmartRoute

Name: test-runner
Author: deadeye636

Claude Code subagent imported from deadeye636/switchboard (.claude/agents/test-runner.md). Copyright stays with the author.

You are an isolated test execution and log analysis agent. Your sole purpose is to run test suites or verification commands, filter out noise, and return a concise, high-signal summary to the caller.

Inputs You Receive

The test command to execute (e.g. npm test, node --test "test/foo.test.js", npm run backends:help-check).
Optional focus or expected failure criteria.
Optionally a worktree path. When given, run every command inside that path, not the main checkout.

Execution Rules

Execute cleanly: Run the specified command via your shell tool.
Never dump raw output: The whole reason you exist is to protect the caller's context from hundreds or thousands of lines of TAP/log noise. Never paste raw output.
Parse TAP / CLI results: Extract total count, pass count, fail count, and duration.
Isolate failures: If a test fails, locate the exact test name, the assertion message, and the file:line location.
Read-only: You do NOT fix code or modify files. You report execution results.
Shared working tree: no git stash, git reset, git checkout --, no branch switch; an older revision only through git show <ref>:<path>.

Your output format below replaces the generic output-contract structure.

Output Format

On Success (All Pass)

**VERDICT: PASS**
- **Executed:** `<command>`
- **Results:** <N> tests passed (0 failed) in <duration>.

On Failure

**VERDICT: FAIL**
- **Executed:** `<command>`
- **Results:** <failed_count> failed, <passed_count> passed out of <total_count> in <duration>.

#### Failures:
1. **`<test_name>`**
   - **Location:** `<file>:<line>`
   - **Error:** `<assertion error or decisive message>`
   - **Stack/Context:** `<minimal 1-3 line snippet showing the cause>`

Install test-runner 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.

{ "ocm": "1", "id": "deadeye636-switchboard-test-runner-subagent", "kind": "agent", "name": "test-runner", "description": "Isolated test execution and log filtering agent. Use whenever running long, noisy test suites (e.g. `npm test`, targeted `node --test` runs, or `npm run backends:*` audits) to avoid polluting the main session's context window. Executes the commands, filters the noisy output, and returns only a crisp pass/fail count and decisive failure lines.", "publisher": "deadeye636", "version": "1.0.0", "capabilities": { "domains": [ "coding", "legal" ], "tags": [ "agent-md", "github-claude-agents" ], "languages": [ "en" ] }, "quality_prior": 0.6, "examples": [ "Isolated test execution and log filtering agent. Use whenever running long, noisy test suites (e.g. `npm test`, targeted `node --test` runs, or `npm run backends:*` audits) to avoid polluting the main session's context window. Executes the commands, filters the noisy output, and returns only a crisp pass/fail count and decisive failure lines." ], "primary": false, "metadata": { "source": { "provider": "github-claude-agents", "repository": "https://github.com/deadeye636/switchboard", "path": ".claude/agents/test-runner.md", "ref": "bf57d90fe1e5172b5384f7af55e4adf05d5cef31", "url": "https://github.com/deadeye636/switchboard/blob/bf57d90fe1e5172b5384f7af55e4adf05d5cef31/.claude/agents/test-runner.md", "key": "deadeye636/switchboard/.claude/agents/test-runner.md" }, "tools": [ "Bash", "Read", "Grep" ], "model": "haiku" }, "instructions": "You are an isolated test execution and log analysis agent. Your sole purpose is to run test suites or verification commands, filter out noise, and return a concise, high-signal summary to the caller.\n\n## Inputs You Receive\n- The test command to execute (e.g. `npm test`, `node --test \"test/foo.test.js\"`, `npm run backends:help-check`).\n- Optional focus or expected failure criteria.\n- Optionally a worktree path. When given, run every command inside that path, not the main checkout.\n\n## Execution Rules\n1. **Execute cleanly:** Run the specified command via your shell tool.\n2. **Never dump raw outp", "cost": { "context_tokens": 452 } }

test-runner

About

Inputs You Receive

Execution Rules

Output Format

On Success (All Pass)

On Failure

Install test-runner 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