Receba novas skills de Design e Frontend toda segunda
1 email curto, só com as skills novas de Design e Frontend. 4 minutos de leitura, sem spam, cancela com 1 clique.
Você confirma o email no primeiro envio. Sem spam. Cancela com 1 clique.
Ler descrição completaRecolher descrição
Implement a story according to the requirements and tasks
Instructions
The story should have tasks grouped by dependencies using the format from story-flow:analyze-task-dependencies skill.
You are an orchestrator, not an implementer.
REQUIRED: Immediately delegate ALL tasks to subagents using the Task tool to run them in isolated context.
FORBIDDEN tools for orchestrator (NEVER use directly):
Glob, Grep, Read
Write, Edit, MultiEdit
WebSearch, WebFetch
TaskCreate, TaskUpdate, TaskGet, TaskList
Task with Explore subagent
Instruct each subagent:
When installing dependencies, use the latest stable version, or a higher patch version within the existing minor if the project pins one.
After implementing UI according to UI design HTML files, verify by taking screenshots of both the UI and the design using mcp__plugin_story-flow_chrome-devtools__navigate_page and mcp__plugin_story-flow_chrome-devtools__take_screenshot.
For every source file created or modified, update the corresponding test file to achieve 100% coverage and ensure all linting warnings and errors are resolved.
One test case per scenario/behavior, not one test per assertion.
Combine test assertions for basic rendering into a single test - only separate tests when testing different states, behaviors, or edge cases.
Every test must have assertions that verify the behavior described in its test name — not just that the code runs without crashing.
No missing assertions: every test must contain at least one expect() that checks a meaningful outcome.
Assertions must be correct: expected values must match the actual business logic, not be copied from implementation output without verification.
Report back to the orchestrator just "completed" or "failed". On success, strictly nothing else. On failure, include a brief summary of the issues encountered.
When a story references the qa-tester subagent for planning BDD scenarios, use the qa-tester subagent and instruct it to load BOTH skills in this order using the Skill tool:
First: story-flow:plan-bdd-scenarios (plugin - general BDD planning protocol)
Then: plan-bdd-scenarios (project-level - overrides/extends the plugin)
Confirm both skills are loaded before continuing with BDD planning.