| name | omg-test |
| description | Run tests via registry-routed omg-tester agent. Compilation checks, coverage reports, failure analysis. Use for 'run tests', 'check coverage', 'why is this test failing'. |
Codex Port Notice
This skill was ported from upstream reference material. Interpret command names, paths, and agent-routing guidance as Codex/Oh My Game Kit equivalents. Prefer active Codex tools and local project instructions over Codex-specific mechanics when they conflict.
Oh My Game Kit Test — Test Runner
Delegate to the registered omg-tester agent. Never ignore failing tests.
Modes
| Flag | Mode | Behavior |
|---|
| (default) | Full | Run entire test suite |
--flaky | Flaky detection | Re-run failing tests up to 3 times, report retry rate per test |
--diff | Diff-aware | Only run tests for files changed since main branch |
--coverage | Coverage | Run with coverage reporting, flag uncovered critical paths |
Flaky Test Detection (--flaky)
- Run test suite normally
- For any failing test, re-run it up to 3 times in isolation
- If test passes on retry: mark as FLAKY (intermittent)
- If test fails all retries: mark as FAILING (genuine failure)
- Report: flaky tests with retry rate, genuine failures separately
Diff-Aware Testing (--diff)
- Run
git diff --name-only origin/main...HEAD to find changed files
- Map changed source files → corresponding test files
- Run only mapped test files
- If no mapping found: run full suite (fallback)
Core Principle
NEVER IGNORE FAILING TESTS. Fix root causes, not symptoms.
Full-Verify Contract — What "Tests Green" Actually Means
"Tests passed" is insufficient evidence to claim done. Three gate types must all pass before reporting complete:
| Gate | Question it answers | Typical command |
|---|
| Type-check | Does it compile? | bun run typecheck / tsc --noEmit / cargo check |
| Lint / Format | Does it meet project style? | bun run lint / biome check / eslint . / ruff check |
| Tests | Does behavior match spec? | bun test / npm test / pytest / cargo test |
These gates are orthogonal. A change can pass tests and fail lint (or vice versa). CI runs all three; claiming "done" after running only one or two is a false positive that surfaces as a red PR later.
Rule
Before claiming work complete:
- Run the kit's type-check script — read the output
- Run the kit's lint/format script — read the output
- Run the kit's test command — read the output
- Only then claim "done" / "green" / "verified"
If the kit has no lint script: note this explicitly ("no lint gate in this project") so it's visible rather than silently skipped.
Common scripts to discover
Check package.json:
jq -r '.scripts | to_entries[] | select(.key | test("^(test|lint|typecheck|check|format|build)$")) | "\(.key): \(.value)"' package.json
If a repo has lint + typecheck + test scripts, run all three. Don't assume one covers the others.
Self-Teach on CI Gate Failures — MANDATORY
Every time a CI quality gate (lint, format, typecheck, test, build, security scan) fails on a PR, you MUST update the responsible skill to prevent the same failure in future sessions. This is the development-principles.md "Update Skills After Every Error" rule applied specifically to CI.
Failure → lesson loop:
- CI gate fails on PR
- Identify which local-verify command would have caught it (the CI step's name + what it runs)
- If that command is NOT in the skill's verify checklist → add it
- If it IS in the checklist but was skipped → note why, reinforce the checklist
- Commit the skill update in the SAME session as the CI fix, not "later"
If you are about to push a fix-up commit for a CI failure without updating a skill, stop and update the skill first. The fix-up without the skill update guarantees the same person (or future you) will hit the same gate again.
Bug trail (why this matters)
PR oh-my-game-kit-cli#79 (2026-04-21) had bun run typecheck clean + 433 tests green, but CI's Lint stage failed on 3 OS matrices because biome check wasn't in the local verify step. Cost: one extra round trip. Had a full-verify contract been followed, the formatter violations would have been caught and fixed locally before push. This skill section exists BECAUSE that happened.
Agent Routing
Follow protocol: .agents/skills/omg-cook/references/routing-protocol.md
This command uses role: omg-tester
Skill Activation
Follow protocol: .agents/skills/omg-cook/references/activation-protocol.md
Workflow
- Compilation check (read console or build output)
- Run tests via registered omg-tester agent
- Analyze test results for failures
- If failures → spawn registered
omg-debugger for root cause
- Report structured results
Module Context for Tester (if installedModules or modules present in metadata.json)
Follow protocol: .agents/skills/omg-cook/references/subagent-injection-protocol.md
Before spawning omg-tester agent, inject:
- Which module's files are being tested (from
.agents/metadata.json)
- Module's test skills if available
- Boundary: "Test files in module {name} should not test cross-module behavior"
Sub-Agent Fork Hygiene
Sub-agent forking: see .agents/skills/omg-architecture/references/fork-hygiene.md.
