parallel-execution

1# Parallel Execution Patterns
2 
3### When to Load
4 
5- **Trigger**: Multi-agent tasks, concurrent operations, spawning subagents, parallelizing independent work
6- **Skip**: Single-step tasks or sequential workflows with no parallelization opportunity
7 
8## Core Concept
9 
10Parallel execution spawns multiple subagents simultaneously using the Agent tool (named `Task` before Claude Code 2.1.63; `Task` still works as an alias). Subagents run in the background by default, so N tasks run concurrently, dramatically reducing total execution time.
11 
12**Critical Rule**: ALL Agent calls MUST be in a SINGLE assistant message for true parallelism. If the calls are in separate messages, they launch one after another.
13 
14## Execution Protocol
15 
16### Step 1: Identify Parallelizable Tasks
17 
18Before spawning, verify tasks are independent:
19 
20- No task depends on another's output
21- Tasks target different files or concerns
22- Can run simultaneously without conflicts
23 
24### Step 2: Prepare Dynamic Subagent Prompts
25 
26Each subagent receives a custom prompt defining its role:
27 
28```
29You are a [ROLE] specialist for this specific task.
30 
31Task: [CLEAR DESCRIPTION]
32 
33Context:
34[RELEVANT CONTEXT ABOUT THE CODEBASE/PROJECT]
35 
36Files to work with:
37[SPECIFIC FILES OR PATTERNS]
38 
39Output format:
40[EXPECTED OUTPUT STRUCTURE]
41 
42Focus areas:
43- [PRIORITY 1]
44- [PRIORITY 2]
45```
46 
47### Step 3: Launch All Tasks in ONE Message
48 
49**CRITICAL**: Make ALL Agent calls in the SAME assistant message:
50 
51```
52I'm launching N parallel subagents:
53 
54[Agent 1]
55…