菜单
Fix all failing tests by running the test suite and spawning subagents to fix each failing test file. Use when the user asks to fix all tests, fix failing tests, or update tests to match current code.
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
基于 SOC 职业分类
Create a demo user and company on local Panda via curl. Authenticates against the local dev server, submits the demo account form, and returns login credentials. Use when the user asks to create a local user, demo account, demo company, or needs a test account on local Panda.
Publish implementation plans to Notion instead of local markdown. Use whenever creating an implementation plan, architecture plan, spike plan, migration plan, or any planning document. Replaces local docs/ output with a Notion page under the Planning docs page.
List and remove old git worktrees sorted by last activity. Shows when each worktree was last touched and lets the user pick which ones to remove. Also cleans up the associated local branch and AVD emulator if one exists. Use when the user says "cleanup worktrees", "remove worktrees", "delete old worktrees", "clean up branches", or "list worktrees".
Delete all Android emulators created by the build-android-app skill. Finds AVDs matching the {project}_{branch} naming convention, shuts down any that are running, and deletes them. Use when the user says "cleanup emulators", "delete emulators", "remove avds", or "clean up android emulators".
Build, install, and launch an Android app on an emulator after code changes. Auto-detects Gradle wrapper, build variant, application ID, and emulator. Use after making code changes to verify compilation, when the user says "build the app", "run the app", "launch on emulator", or when a workspace rule says to build after changes.
Build, install, and launch an iOS app on the Simulator after code changes. Auto-detects project file, scheme, bundle ID, and simulator. Use after making code changes to verify compilation, when the user says "build the app", "run the app", "launch on simulator", or when a workspace rule says to build after changes.
| name | fix-all-tests |
| description | Fix all failing tests by running the test suite and spawning subagents to fix each failing test file. Use when the user asks to fix all tests, fix failing tests, or update tests to match current code. |
Fix all failing tests by running the full test suite, then spawning parallel subagents to fix each failing test file.
Key assumption: The implementation code is correct. Tests need to be updated to match current behavior.
Run all tests and capture output:
npm run test 2>&1
Parse the output to identify all failing test files.
From the test output, identify each unique test file that has failures. Look for patterns like:
FAIL src/path/to/test.spec.ts● Test Suite: path/to/testCreate a list of all failing test file paths.
For each failing test file, spawn a subagent using the Task tool with these parameters:
subagent_type: "generalPurpose"
description: "Fix tests in [filename]"
prompt: |
Fix the failing tests in [full file path].
IMPORTANT: Assume the implementation code is CORRECT. The tests need to be updated to match the current behavior.
Steps:
1. Read the failing test file
2. Read the implementation file(s) being tested
3. Understand what the code actually does
4. Update the test expectations to match the actual behavior
5. Run the specific test file to verify: npm run test -- [file path]
6. Repeat until all tests in this file pass
7. Lint the file: npm run lint -- [file path] --fix
8. Fix any remaining lint errors
Do NOT modify implementation code - only update tests.
Parallelization: Launch up to 4 subagents concurrently. If more than 4 files are failing, wait for some to complete before launching more.
After all subagents complete, run the full test suite again:
npm run test
If any tests still fail, repeat Step 3 for remaining failures.
For a file src/features/auth/auth.spec.ts:
Fix the failing tests in src/features/auth/auth.spec.ts.
IMPORTANT: Assume the implementation code is CORRECT. The tests need to be updated to match the current behavior.
Steps:
1. Read the failing test file
2. Read the implementation file(s) being tested
3. Understand what the code actually does
4. Update the test expectations to match the actual behavior
5. Run the specific test file to verify: npm run test -- src/features/auth/auth.spec.ts
6. Repeat until all tests in this file pass
7. Lint the file: npm run lint -- src/features/auth/auth.spec.ts --fix
8. Fix any remaining lint errors
Do NOT modify implementation code - only update tests.