swarm

Orchestrates parallel cloud workers to cover slices, race briefs, and aggregate one report.

136|8|Updated May 9, 2026
One-click install
npx skills add https://github.com/Sma1lboy/rove --skill swarm-sma1lboy
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: swarm
Source: https://github.com/Sma1lboy/rove/tree/main/.agents/skills/pstack/skills/swarm
Command: npx skills add https://github.com/Sma1lboy/rove --skill swarm-sma1lboy

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Coordinating many parallel AI workers by hand is error-prone: briefs get inconsistent, results arrive scattered, and aggregation becomes manual guesswork. This Skill standardizes fan-out, monitoring, and consolidation so parallel work returns a single consolidated report. ## Core Features & Use Cases - Structured Fan-Out: Spawns N background cloud workers in one message, each with a standalone brief covering goal, scope, verification, and reporting format. - Flexible Shapes: Supports coverage (partitioned slices), races (identical briefs with first-pass, rank-all, or best-of selection), and mixed modes. - Aggregation & Reporting: Collects terminal results into a compact table with evidenced issues, explicit gaps or dropouts, and one consolidated in-chat report. - Use Case: Ask to swarm a codebase review across five modules; each worker audits its slice in an isolated worktree, and you receive one report with PASS/ISSUES/BLOCKED status per module. ## Quick Start Ask the agent to swarm this task with five parallel workers and return one consolidated report.

Frequently Asked Questions about swarm

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I run parallel AI workers on one task?▼

Use the swarm workflow: frame a done predicate, choose a shape (slices, race, or mixed), then spawn all N workers in one message as background cloud agents. Each worker gets a standalone brief and its own writable output location.

How do I choose between coverage and race mode for parallel agents?▼

Use coverage when work partitions cleanly into slices that each need a result. Use a race when workers run identical briefs and you select by first pass, rank all, or best-of, declared before spawning.

Can swarm workers use different models or branches?▼

Workers inherit the session model by default; pass a model only for cheaper mechanical work or a deliberate model race. Use cloud_base_branch when a worker must start from a non-default pushed branch.

What happens if a parallel worker drops out or fails?▼

The swarm proceeds with N-1 workers and notes the dropout. The final report includes explicit gaps or dropouts alongside the result table so missing coverage is visible.

When should I not use a parallel worker swarm?▼

Avoid swarming when the task is sequential, when workers would write to shared state without isolation, or when a single worker can complete the job. Local environment is only for workers needing access to the user's machine.