dispatching-parallel-agents

Dispatching Parallel Agents

Dispatching Parallel Agents is an agent skill from Superpowers: when several problems are independent, your agent sends a focused subagent to each at once, then checks the results together.

Checked by CookiesForAI on : license and last change of the skill re-read from GitHub.

Written by

Install Dispatching Parallel Agents

Install only this skill with the skills CLI

Terminal
npx skills add obra/superpowers --skill dispatching-parallel-agents

Or, in Claude Code, install the whole Superpowers plugin from Anthropic's official marketplace

Agent chat
/plugin install superpowers@claude-plugins-official

Or paste this into your coding agent: Install the agent skill dispatching-parallel-agents from github.com/obra/superpowers. Checked by CookiesForAI on : license and last change of the skill re-read from GitHub. A skill can include scripts that run on your computer, so read its source first.

What Dispatching Parallel Agents does

Dispatching Parallel Agents is one of the skills in Superpowers, Jesse Vincent's open-source method for AI coding agents. It loads when there are two or more independent tasks that share no state and do not depend on each other, such as test failures in different files or bugs in separate subsystems. Instead of investigating them one after another, the agent gives each problem its own subagent, and they work at the same time.

In Dispatching Parallel Agents, each subagent gets a prompt the agent writes for it, never the session's own history: one test file or subsystem as its scope, a clear goal, constraints such as not changing other code, and the summary it must return. The skill contrasts vague prompts ("fix all the tests") with focused ones that paste in the error messages and test names. All dispatches go out in the same response, which is what makes them run in parallel.

Dispatching Parallel Agents also says when not to use it: failures that are related, problems that need the whole system in view, exploratory debugging where the cause is still unknown, and work where agents would edit the same files. When the subagents return, the agent reads each summary, checks that the fixes do not conflict, runs the full test suite, and spot-checks the changes, since agents can make systematic errors.

When to use Dispatching Parallel Agents

  • You have several failing tests in different files that look unrelated.
  • You have separate parts of your project that broke at the same time and want them fixed in parallel, not one by one.
  • You want your main agent's context kept free for coordination while other agents do the digging.

When to pick something else

  • The failures are related or the cause is unknown: use Systematic Debugging.
  • You are carrying out a written plan task by task: use Subagent-Driven Development.

Which agents Dispatching Parallel Agents works in

Jesse Vincent documents Dispatching Parallel Agents for Claude Code, Codex (app and CLI), Cursor, Gemini CLI, GitHub Copilot CLI, OpenCode, Antigravity, Devin CLI, Factory Droid, Grok Build CLI, Kimi Code, Pi, Qwen Code, Hermes Agent, Muse. The open skills CLI also installs it into 78 agents, including Claude Code, Codex, Cursor, Gemini CLI, GitHub Copilot, OpenCode (we listed it with the CLI on ). See where each agent looks for skills.

Superpowers is written in Claude Code's vocabulary (subagent dispatch, todos, the Skill tool) and ships tool mappings for Codex, Gemini CLI, Pi, Antigravity, Hermes Agent and Muse. Installed as the full plugin, a session-start hook loads Using Superpowers so the agent reaches for skills on its own; one skill installed with the skills CLI gets no hook and may point to sibling skills you do not have. This skill needs a subagent tool: Claude Code's Agent tool, or in Codex the multi-agent feature switched on in config.toml (multi_agent = true).

Dispatching Parallel Agents license

Dispatching Parallel Agents is published under MIT.

More from Jesse Vincent

Lists that include dispatching-parallel-agents