| name | ci-run |
| description | Execute android-wifi-mcp test cases against the attached Android device.
Use when the user wants to run tests, execute test cases, or verify MCP tool behavior end-to-end.
|
| disable-model-invocation | true |
Run android-wifi-mcp Test Cases
Execute YAML test cases against the attached Android device using the MCP stdio transport.
{{input}}
PURPOSE
Run YAML test cases by spawning the MCP server over stdio, calling tools, and evaluating results.
AGENT WORKFLOW
Step 1: Identify the scope
Input can be:
- A test ID (e.g.,
TC-SMK-003) — run that specific test
- A suite name (e.g.,
smoke, wifi) — run the whole suite
- A tag (e.g.,
device, network) — run all tests tagged with it
- Empty — run everything
Step 2: Pre-flight checks
Before running:
adb devices lists at least one device-state entry
dist/index.js exists at the repo root (run npm run build if not)
cicd/tests/node_modules exists (run cd cicd/tests && npm install if not)
If any of those is missing, fix it and report what you did.
Step 3: Execute
From the repo root:
cd cicd/tests && npx tsx src/cli.ts run --suite <suite>
cd cicd/tests && npx tsx src/cli.ts run --id <TC-XXX-NNN>
cd cicd/tests && npx tsx src/cli.ts run --tag <tag>
The runner streams progress to stderr and writes JSON results to cicd/results/<timestamp>_<suite>/.
Per-test, the executor snapshots WiFi state before and restores after — even on failure — so a flaky test doesn't poison subsequent ones.
Step 4: Interpret results
A non-zero exit code from the runner means at least one test failed. The console reporter prints a summary table; per-test JSON reports are in the results directory for deeper inspection.
For a failing test:
- Open
cicd/results/<timestamp>/<test-id>.json for the full step-by-step trace
- Common causes:
- Tool error → check the
stdout field for the actual MCP response
- Pattern miss → confirm you used bare strings (see ci-testcase skill for the gotcha)
- Device state drift → confirm restore actually ran (look for
Restored device state in the log)
Step 5: Report
Output:
- Pass/fail count
- For failures: test ID, the failing step, and the first diagnostic line
- Path to the results directory
OUTPUT
Test results summary plus the path to the JSON results for the run.