Diablo_Rain/Diablo_ClaudeMD_Ricing_example
1
0
Fork 0
Code Issues Pull Requests Actions 1 Packages Projects Releases Wiki Activity
8b47d8e2b8
Diablo_ClaudeMD_Ricing_example/skills/swarm/SKILL.md
diablo 50fa79407d
Some checks are pending
CI — CoM Config Validation / Validate JSON Configs (push) Waiting to run
Details
CI — CoM Config Validation / Validate YAML Configs (push) Waiting to run
Details
CI — CoM Config Validation / Lint Shell Scripts (push) Waiting to run
Details
CI — CoM Config Validation / Secret Detection (push) Waiting to run
Details
CI — CoM Config Validation / Lint Markdown (push) Waiting to run
Details
CI — CoM Config Validation / Validate CODEOWNERS (push) Waiting to run
Details
CoM Claude Command Center — sanitized public configuration 
Public, sanitized mirror of an AI orchestration command center: agents, skills,
MCP servers, slash-command workflows. All infrastructure identifiers, hostnames,
mesh IPs/subnets, repo paths, maintainer identity, and hardware fleet specifics
scrubbed to <placeholders>; session debug logs and host-specific memory removed.
No live credentials. Verified clean by automated leak sweep. See SANITIZATION.md.

churchofmalware.org . authorized research only
2026-06-10 02:02:03 -04:00

1.6 KiB
Raw Blame History

name description argument-hint allowed-tools
swarm Parallel agent orchestration. Decomposes large tasks into waves, spawns specialist agents from the A2A registry concurrently, collects and synthesizes results. Enterprise-scale execution mode.
task description
Bash, Read, Grep, Glob, Agent, WebFetch, WebSearch

/swarm — CoM Parallel Agent Orchestration

Decomposes a large task and spawns specialist agents from the A2A registry in coordinated waves.

Step 1: Registry Load

Read ~/.claude/a2a/agent-cards.json — parse agents, capabilities, hardware limits.

Step 2: Task Analysis

Parse $ARGUMENTS. Use Sequential Thinking to decompose:

  1. Overarching goal
  2. Independent sub-tasks (can run in parallel)
  3. Dependent sub-tasks (must wait)
  4. Agent mapping by capabilities keywords
  5. Group into waves respecting max_concurrent_agents limit

Step 3: Wave Planning

Present formatted plan with waves, agent assignments, deliverables, and hardware constraints.

Step 4: Conflict Prevention

  • No two agents write the same file (serialize if needed)
  • Cipher and Vanguard never run concurrently on same crate
  • Aegis and Specter can run in parallel (both read-only)
  • Total concurrent <= hardware profile limit (reserve 1 slot for overhead)

Step 5: Approval and Execution

Ask: "Deploy swarm? [y/N]" On approval: execute waves sequentially, agents within each wave in parallel.

Step 6: Post-Swarm Report

Show results per agent, failures, and final unified deliverable.

CONSTRAINTS

  • Respect hardware limits
  • If task is inherently sequential, recommend /go instead
  • Every swarm requires user approval
  • Failed agents don't block others