ue-test-authoring

name	ue-test-authoring
description	Use when writing or modifying UE automated tests (Automation, CQTest, Functional, Gauntlet, LowLevel) with Rider MCP available. Value over bash/grep: IDE diagnostics catch test registration errors, wrong RunTest return type, and missing includes before a build; get_symbol_info verifies the API under test. DO NOT TRIGGER for: debugging existing test failures (use ue-live-debugging), Blueprint-only testing. When Rider MCP is unavailable, runs in reduced mode — standard file tools only, IDE diagnostics skipped.
allowed-tools	Read Glob Grep Bash Write Edit ToolSearch
metadata	{"argument-hint":"[test type, feature/system to test, or failing scenario]"}

UE Test Authoring

End-to-end workflow for writing Unreal Engine automated tests backed by Rider MCP for IDE-grade code quality.
Three additions over plain test writing: (1) Rider diagnostics catch issues before a full build, (2) lint_files enforces consistency across test files, (3) get_project_problems surfaces cross-file issues invisible to grep.

---

GATE — mandatory checks before any test code is written

1. UE Project Check

Verify the current working directory contains a .uproject file:

find . -maxdepth 1 -name "*.uproject" | head -1

If no .uproject found → STOP.

"This skill requires an Unreal Engine project (a .uproject file must be in the working directory). The current directory does not appear to be a UE project. Navigate to the project root and retry."

2. Task Is Test-Authoring Check

The task must involve writing or modifying test code. If the user is asking about something that has no test authoring component — e.g. fixing a runtime bug, building the project, or changing non-test source files — clarify scope before proceeding.

If there is no test to write or modify → STOP and ask the user what test they want authored.

3. Rider MCP Availability Check

Check the <system-reminder> deferred-tool list for Rider MCP tools. Load live schemas with ToolSearch before calling any tool. Schemas are authoritative for parameter names — never guess. If execute_tool is the only tool returned, use CLI mode (see reference/rider-mcp-tools.md — execute_tool mode).

If no Rider MCP tools appear in the deferred list:

"Rider MCP tools are unavailable. Open Rider with this project loaded and the MCP server enabled, then retry. Falling back to standard file tools — IDE diagnostics will not run."

Proceed with standard tools (Read/Write/Edit/Grep/Bash) only, skipping all mcp__<prefix>__* steps. Document that quality steps were skipped.

---

Path Selection

Fast path for adding a single test case to an existing test file — no new module, no new framework, no Build.cs changes:

1. Verify .uproject (Gate 1)
2. Locate the existing test file with rg/Grep or Rider search_text
3. Read 1–2 existing test cases in the same file
4. Add the minimal test case using the standard Edit tool
5. Run get_file_problems on the changed file if Rider MCP is available
6. Show git diff

Full workflow when creating a new test module, adding a new framework, changing Build.cs, or modifying multiple test files → continue to Checklist below.

Checklist

Use the agent's native planning/todo mechanism when available. For simple one-file tasks, keep the plan implicit and proceed directly. For complex changes, track:

1. GATE — UE project check + test task check + Rider prefix resolution
2. Clarify — ask targeted questions if the framework or scope is ambiguous (skip if clear)
3. Select framework — choose the right test framework using the decision guide below
4. Pre-flight — read .uproject, locate existing test modules, scan patterns via Rider
5. Write test — create or modify test .h/.cpp using framework patterns
6. Rider diagnostics — get_file_problems per changed file; fix all errors and warnings
7. Batch lint (if multiple test files changed or new patterns introduced) — lint_files; fix remaining issues
8. Build — build_solution_start; poll build_solution_state until done; fix errors
9. Post-build quality (for non-trivial changes) — get_project_problems; address Critical/Important
10. Reformat — reformat_file on each changed file

---

Workflow

Search routing: use rg/Grep for portable text discovery; use Rider search_text/search_file when IDE index, generated/reflected UE code, unsaved editor state, or result compactness is likely to help. Use Rider semantic tools (search_symbol, get_file_problems, lint_files, build, get_project_problems) for code intelligence.

Step 0 — Clarify (if ambiguous)

Skip if the request names a specific framework, class, or test scenario.

Ask one question at a time, max two questions total:

• "Is this a pure logic test (no actors, no world) or does it need an in-game context?" — determines Automation/CQTest vs Functional Test
• "Does the test need to verify replication or server/client behavior?" — affects CQTest PIENetworkComponent vs plain test
• "Is there an existing test module I should add to, or should this be a new module?" — check with list_directory_tree before asking

Step 1 — Select Framework

Pick the minimal framework that covers the testing need. See reference/ue-test-patterns.md — Framework Selection for the decision matrix.

Step 2 — Pre-flight

Read project context before writing any test: .uproject file and Source/<TestModule>/<TestModule>.Build.cs using the standard Read tool.

Find existing test modules and patterns using search_text (look for IMPLEMENT_SIMPLE_AUTOMATION_TEST, TEST_CLASS, DEFINE_SPEC). Use Glob or list_directory_tree to browse directory structure.

Use search_symbol to locate the module under test for Build.cs dependency lookup.

Read 1–2 existing test files in the same area using the standard Read tool to match conventions.

Determine from pre-flight: whether a test module exists (add to it vs. create new), which test macros the project uses (match them), and required PrivateDependencyModuleNames.

See reference/ue-test-patterns.md — New Test Module Setup if a new module is needed.

Step 3 — Write Test

Create new test module (if none exists) — see reference/ue-test-patterns.md — New Test Module Setup.

Create a new test file using the standard Write tool.

Modify an existing test file using the standard Edit tool.

See reference/ue-test-patterns.md for ready-to-use patterns for each framework and critical pitfalls.

Step 4 — Rider Diagnostics (per file)

After writing each file, run get_file_problems on it immediately.

Act on every result:

• Error → fix before proceeding; re-run get_file_problems to confirm clear
• Warning → fix unless it's a known intentional pattern (document why if skipping)
• Hint / Info → note for later; don't block on these

Iterate: edit → diagnose → edit until zero errors and zero warnings on all changed files.

Step 5 — Batch Lint (skip for isolated single-file test additions)

If multiple test files changed or new patterns were introduced, run lint_files.

Fix any issues surfaced here that get_file_problems missed (cross-file include violations, project-level style rules).

Step 6 — Build

Compile via Rider using build_solution_start — do NOT shell-build or ask the user to rebuild manually.

Poll using build_solution_state until state != "Running".

On build failure:

• Read the error output from build_solution_state
• Identify which file/line caused the error
• Fix with the standard Edit tool, re-run get_file_problems on the fixed file, then rebuild
• Do NOT proceed to Step 7 with build errors outstanding

Step 7 — Post-Build Quality Gate (skip for small test additions)

For non-trivial changes, run get_project_problems. Filter results to files you changed. For each issue:

• Error → fix immediately
• Warning on your files → fix unless intentionally deferred

Step 8 — Reformat

Use reformat_file on every file you created or modified.

---

see: reference/rider-mcp-tools.md — ALL Rider MCP tools: complete parameter reference, execute_tool mode table, UE editor/asset/debugger tools see: reference/rider-tools.md — Test authoring workflow patterns: fix-loop, common lookups, test discovery see: reference/ue-test-patterns.md — Framework patterns (Automation, CQTest, Functional, Gauntlet, LowLevel), new module setup, critical pitfalls