dispatching-parallel-agents

Name: dispatching-parallel-agents
Author: aaddrick

120stars16forks

How to install dispatching-parallel-agents?

aaddrick/claude-pipeline/dispatching-parallel-agents

$npx -y skills add aaddrick/claude-pipeline --skill dispatching-parallel-agents

Installs into the current project.

Summary

Use when facing 2+ independent tasks that can be worked on without shared state or sequential dependencies

Files · 1

View on GitHub

SKILL.md

1	# Dispatching Parallel Agents
2
3	## Overview
4
5	When you have multiple unrelated failures (different test files, different subsystems, different bugs), investigating them sequentially wastes time. Each investigation is independent and can happen in parallel.
6
7	Core principle: Dispatch one agent per independent problem domain. Let them work concurrently.
8
9	## When to Use
10
11	```dot
12	digraph when_to_use {
13	"Multiple failures?" [shape=diamond];
14	"Are they independent?" [shape=diamond];
15	"Single agent investigates all" [shape=box];
16	"One agent per problem domain" [shape=box];
17	"Can they work in parallel?" [shape=diamond];
18	"Sequential agents" [shape=box];
19	"Parallel dispatch" [shape=box];
20
21	"Multiple failures?" -> "Are they independent?" [label="yes"];
22	"Are they independent?" -> "Single agent investigates all" [label="no - related"];
23	"Are they independent?" -> "Can they work in parallel?" [label="yes"];
24	"Can they work in parallel?" -> "Parallel dispatch" [label="yes"];
25	"Can they work in parallel?" -> "Sequential agents" [label="no - shared state"];
26	}
27	```
28
29	Use when:
30	- 3+ test files failing with different root causes
31	- Multiple subsystems broken independently
32	- Each problem can be understood without context from others
33	- No shared state between investigations
34
35	Don't use when:
36	- Failures are related (fix one might fix others)
37	- Need to understand full system state
38	- Agents would interfere with each other
39
40	## The Pattern
41
42	### 1. Identify Independent Domains
43
44	Group failures by what's broken:
45	- File A tests: Tool approval flow
46	- File B tests: Batch completion behavior
47	- File C tests: Abort functionality
48
49	Each domain is independent - fixing tool approval doesn't affect abort tests.
50
51	### 2. Create Focused Agent Tasks
52
53	Each agent gets:
54	- Specific scope: One test file or subsystem
55	- Clear goal: Make these tests pass
56	- Constraints: Don't change other code
57	- Expected output: Summary of what you found and fixed
58
59	### 3. Dispatch in Parallel
60
61	```typescript
62	// In Claude Code / AI environment
63	Task("Fix agent-tool-abort.test.ts failures")
64	Task("Fix batch-completion-behavior.test.ts failures")
65	Task("Fix tool-approval-race-conditions.test.ts failures")
66	// All three run concurrently
67	```
68
69	### 4. Review and Integrate
70
71	When agents return:
72	- Read each summary
73	- Verify fixes don't conflict
74	- Run full test suite
75	- Integrate all changes
76
77	## Agent Prompt Structure
78
79	Good agent prompts are:
80	1. Focused - One clear problem domain
81	2. Self-contained - All context needed to understand the problem
82	3. Specific about output - What should the agent return?
83
84	```markdown
85	Fix the 3 failing tests in src/agents/agent-tool-abort.test.ts:
86
87	1. "should abort tool with partial output capture" - expects 'interrupted at' in message
88	2. "should handle mixed completed and aborted tools" - fast tool aborted instead of completed
89	3. "should properly track pendingToolCount" - expects 3 results but gets 0
90
91	These are timing/race condition issues. Your task:
92
93	1. Read the test file and understand what each test verifies
94	2. Identify root cause - timing issues or actual bugs?
95	3. Fix by:
96	- Replacing arbitrary timeouts with event-based waiting
97	- Fixing bugs in abort implementation if found
98	- Adjusting test expectations if testing changed behavior
99
100	Do NOT just increase timeouts - find the real issue.
101
102	Return: Summary of what you found and what you fixed.
103	```
104
105	## Common Mistakes
106
107	❌ Too broad: "Fix all the tests" - agent gets lost
108	✅ Specific: "Fix agent-tool-abort.test.ts" - focused scope
109
110	❌ No context: "Fix the race condition" - agent doesn't know where
111	✅ Context: Paste the error messages and test names
112
113	❌ No constraints: Agent might refactor everything
114	✅ Constraints: "Do NOT change production code" or "Fix tests only"
115
116	❌ Vague output: "Fix it" - you don't know what changed
117	✅ Specific: "Return summary of root cause and changes"
118
119	## When NOT to Use
120
121	Related failures: Fixing one might fix others - investigate together first
122	Need full context: Understanding requires seeing entire system
123	Exploratory debugging: You don't know what's broken yet
124	Shared state: Agents would interfere (editing same files, using same resources)
125
126	## Real Example from Session
127
128	Scenario: 6 test failures across 3 files after major refactoring
129
130	Failures:
131	- agent-tool-abort.test.ts: 3 failures (timing issues)
132	- batch-completion-behavior.test.ts: 2 failures (tools not executing)
133	- tool-approval-race-conditions.test.ts: 1 failure (execution count = 0)
134
135	Decision: Independent domains - abort logic separate from batch completion separate from race conditions
136
137	Dispatch:
138	```
139	Agent 1 → Fix agent-tool-abort.test.ts
140	Agent 2 → Fix batch-completion-behavior.test.ts
141	Agent 3 → Fix tool-approval-race-conditions.test.ts
142	```
143
144	**

People who liked this also viewed

6 picks

Type

paperclip-create-agentCreate new agents in Paperclip with governance-aware hiring. Use when you need to inspect adapter configuration options, compare existing agent configs, draft a new agent prompt/config, and submit a hSkillsJul 2026157k5
subagent-driven-developmentUse when executing implementation plans with independent tasks in the current sessionSkillsJul 2026154k259k
dispatching-parallel-agentsUse when facing 2+ independent tasks that can be worked on without shared state or sequential dependenciesSkillsJul 2026143k259k
browser-act-skill-forgeForges reusable Skill packages (SKILL.md + scripts) from website exploration via browser-act — no re-exploration later. Use when: user wants a reusable Skill for any website, needs to understand a sitSkillsJul 202685k4.6k
wizardGenerate an interactive bash wizard that walks a human through a manual procedure — third-party setup, a one-off migration, an A→B state transition — opening URLs, capturing values, confirming each stSkillsJul 202675k182k
wonda-cliUsing the Wonda CLI to generate images, videos, music, and audio from the terminal — plus LinkedIn, Reddit, and X/Twitter research and automationSkillsJul 202666k137

dispatching-parallel-agents

120stars16forks

View on GitHub

How to install dispatching-parallel-agents?

aaddrick/claude-pipeline/dispatching-parallel-agents

$npx -y skills add aaddrick/claude-pipeline --skill dispatching-parallel-agents

Installs into the current project.

Summary

Use when facing 2+ independent tasks that can be worked on without shared state or sequential dependencies

Files · 1

View on GitHub

SKILL.md

1	# Dispatching Parallel Agents
2
3	## Overview
4
5	When you have multiple unrelated failures (different test files, different subsystems, different bugs), investigating them sequentially wastes time. Each investigation is independent and can happen in parallel.
6
7	Core principle: Dispatch one agent per independent problem domain. Let them work concurrently.
8
9	## When to Use
10
11	```dot
12	digraph when_to_use {
13	"Multiple failures?" [shape=diamond];
14	"Are they independent?" [shape=diamond];
15	"Single agent investigates all" [shape=box];
16	"One agent per problem domain" [shape=box];
17	"Can they work in parallel?" [shape=diamond];
18	"Sequential agents" [shape=box];
19	"Parallel dispatch" [shape=box];
20
21	"Multiple failures?" -> "Are they independent?" [label="yes"];
22	"Are they independent?" -> "Single agent investigates all" [label="no - related"];
23	"Are they independent?" -> "Can they work in parallel?" [label="yes"];
24	"Can they work in parallel?" -> "Parallel dispatch" [label="yes"];
25	"Can they work in parallel?" -> "Sequential agents" [label="no - shared state"];
26	}
27	```
28
29	Use when:
30	- 3+ test files failing with different root causes
31	- Multiple subsystems broken independently
32	- Each problem can be understood without context from others
33	- No shared state between investigations
34
35	Don't use when:
36	- Failures are related (fix one might fix others)
37	- Need to understand full system state
38	- Agents would interfere with each other
39
40	## The Pattern
41
42	### 1. Identify Independent Domains
43
44	Group failures by what's broken:
45	- File A tests: Tool approval flow
46	- File B tests: Batch completion behavior
47	- File C tests: Abort functionality
48
49	Each domain is independent - fixing tool approval doesn't affect abort tests.
50
51	### 2. Create Focused Agent Tasks
52
53	Each agent gets:
54	- Specific scope: One test file or subsystem
55	- Clear goal: Make these tests pass
56	- Constraints: Don't change other code
57	- Expected output: Summary of what you found and fixed
58
59	### 3. Dispatch in Parallel
60
61	```typescript
62	// In Claude Code / AI environment
63	Task("Fix agent-tool-abort.test.ts failures")
64	Task("Fix batch-completion-behavior.test.ts failures")
65	Task("Fix tool-approval-race-conditions.test.ts failures")
66	// All three run concurrently
67	```
68
69	### 4. Review and Integrate
70
71	When agents return:
72	- Read each summary
73	- Verify fixes don't conflict
74	- Run full test suite
75	- Integrate all changes
76
77	## Agent Prompt Structure
78
79	Good agent prompts are:
80	1. Focused - One clear problem domain
81	2. Self-contained - All context needed to understand the problem
82	3. Specific about output - What should the agent return?
83
84	```markdown
85	Fix the 3 failing tests in src/agents/agent-tool-abort.test.ts:
86
87	1. "should abort tool with partial output capture" - expects 'interrupted at' in message
88	2. "should handle mixed completed and aborted tools" - fast tool aborted instead of completed
89	3. "should properly track pendingToolCount" - expects 3 results but gets 0
90
91	These are timing/race condition issues. Your task:
92
93	1. Read the test file and understand what each test verifies
94	2. Identify root cause - timing issues or actual bugs?
95	3. Fix by:
96	- Replacing arbitrary timeouts with event-based waiting
97	- Fixing bugs in abort implementation if found
98	- Adjusting test expectations if testing changed behavior
99
100	Do NOT just increase timeouts - find the real issue.
101
102	Return: Summary of what you found and what you fixed.
103	```
104
105	## Common Mistakes
106
107	❌ Too broad: "Fix all the tests" - agent gets lost
108	✅ Specific: "Fix agent-tool-abort.test.ts" - focused scope
109
110	❌ No context: "Fix the race condition" - agent doesn't know where
111	✅ Context: Paste the error messages and test names
112
113	❌ No constraints: Agent might refactor everything
114	✅ Constraints: "Do NOT change production code" or "Fix tests only"
115
116	❌ Vague output: "Fix it" - you don't know what changed
117	✅ Specific: "Return summary of root cause and changes"
118
119	## When NOT to Use
120
121	Related failures: Fixing one might fix others - investigate together first
122	Need full context: Understanding requires seeing entire system
123	Exploratory debugging: You don't know what's broken yet
124	Shared state: Agents would interfere (editing same files, using same resources)
125
126	## Real Example from Session
127
128	Scenario: 6 test failures across 3 files after major refactoring
129
130	Failures:
131	- agent-tool-abort.test.ts: 3 failures (timing issues)
132	- batch-completion-behavior.test.ts: 2 failures (tools not executing)
133	- tool-approval-race-conditions.test.ts: 1 failure (execution count = 0)
134
135	Decision: Independent domains - abort logic separate from batch completion separate from race conditions
136
137	Dispatch:
138	```
139	Agent 1 → Fix agent-tool-abort.test.ts
140	Agent 2 → Fix batch-completion-behavior.test.ts
141	Agent 3 → Fix tool-approval-race-conditions.test.ts
142	```
143
144	**

People who liked this also viewed

6 picks

Type