Orchestrates parallel cloud workers to cover slices or race briefs and returns one aggregated report.
npx skills add https://github.com/mmdmcy/fluttAIrbar --skill swarm-mmdmcy
Please help me install this Agent Skill. Skill: swarm Source: https://github.com/mmdmcy/fluttAIrbar/tree/main/plugins/pstack/skills/swarm Command: npx skills add https://github.com/mmdmcy/fluttAIrbar --skill swarm-mmdmcy
High-intent search queries and answers about installing and using this skill.
Use the swarm workflow: frame a done predicate, choose a coverage or race shape, then spawn N background workers in one message with standalone briefs. Each worker reports PASS, ISSUES, or BLOCKED with evidence, and the parent aggregates everything into one report.
Coverage partitions the task into distinct slices so each worker handles a different scope. A race gives N workers identical briefs and applies a declared selection rule such as first pass, rank all, or best-of to pick the winning result.
Yes. The worker model is picked from the swarm workers entry in ~/.codex/pstack-models.md when present, otherwise it defaults to auto. For a model race, each arm's model must be named up front before spawning.
The workflow proceeds with N-1 workers and explicitly notes the dropout in the final report. Aggregation still requires every required slice to have a result for coverage shapes, and gaps are listed alongside the result table.
Each worker that writes gets its own writable output, such as a dedicated worktree, branch, or a directory like /tmp/swarm-<slug>/worker-<n>/. Workers needing a non-default starting point receive a cloud_base_branch parameter.
adenhqOrchestrates parallel worker agents over tracker tables using deterministic convergence playbooks.
farzanmrzTriage feature-branch findings and execute the final ship gate.
ToanPV90Run a revert-and-confirm protocol to verify regression tests catch defects.
doanchienthangdevOrchestrates concurrent specialized agents for parallel task execution and multi-review workflows.
