parallel-workflows_skill

Overview

This skill optimizes running multiple independent tasks in parallel to save time and improve throughput. It detects which tasks can be executed concurrently, launches background Task agents, and aggregates results into a concise report. Use it when you want safe, efficient parallel execution without breaking task dependencies.

How this skill works

The skill inspects a list of planned tasks and applies simple rules to determine parallelizability: no shared file edits, no data dependencies, and no required execution order. It starts eligible tasks with run_in_background, monitors progress, collects TaskOutput, and produces a unified execution report with timings, status, and changed files. Failed tasks are surfaced with error details and provide retry options.

When to use it

• You ask to run multiple actions together ("run in parallel", "do these together").
• You have multiple independent file edits or analyses that do not touch the same files.
• You want to run lint/type-check/test or similar checks concurrently to speed CI-style checks.
• You want to implement independent features or components in parallel.
• You need a consolidated report of several simultaneous jobs' outputs and file changes.

Best practices

• Confirm tasks are independent before parallelizing: avoid same-file edits or sequential data handoffs.
• Keep tasks small and clearly scoped so the skill can accurately judge independence.
• Reserve parallel runs for CPU/IO-tolerant workflows like analysis, linting, testing, or independent component work.
• Collect and review the generated integration report to capture failed tasks and changed files.
• If a task depends on another's output, run them sequentially or create an explicit dependency step.

Example use cases

• Project structure analysis: run directory analysis, dependency checks, and quality scans at once to produce a single project summary.
• CI-style checks: execute lint, type-check, and test suites in parallel, then run build after they succeed.
• Independent UI work: create header, footer, and sidebar components in parallel agents and merge results.
• Bulk code audits: parallelize file-by-file static analysis across source directories to reduce total scan time.
• Partial failure handling: run 5 tasks concurrently, keep successful outputs, show error details for failing ones, and offer re-run.