Help us improve
Share bugs, ideas, or general feedback.
Share bugs, ideas, or general feedback.
Share bugs, ideas, or general feedback.
By himkt
Orchestrate multi-agent teams for research, design documents, and presentations using a message broker CLI with structured workflows, quality loops, and automated chart generation.
npx claudepluginhub himkt/cafleet --plugin cafleetActive monitoring mechanism for CAFleet Directors. Documents the cron-like loop primitive per backend (Claude Code: CronCreate + /loop; codex: no in-session scheduling, fallback options listed) and the team-facilitation instructions (poll, ACK, dispatch queued work, health-check, escalate). Load whenever you are about to spawn or manage CAFleet team members. Foundation layer — load before agent-team-supervision.
Governance layer for CAFleet Directors. Loads agent-team-monitoring as a hard prerequisite. Defines Core Principle, Communication Model, Idle Semantics, Authorization-Scope Guard, Spawn Protocol, User Delegation, Stall Response (cross-reference), and Cleanup. Load both monitoring and supervision whenever you are about to spawn or manage CAFleet team members (any 'cafleet member create' call).
Resolve the base directory for output files. Loaded by consuming skills via Skill(cafleet:base-dir). Do NOT invoke directly.
Interact with the CAFleet message broker. Use when an agent needs to register, send/receive messages, poll inbox, acknowledge messages, or discover other agents.
Create data visualizations and charts using matplotlib. Triggered when user asks to create a chart, plot, graph, or visualize data. Also invokable via /cafleet:create-figure. Do NOT use plt.show() — always save to PNG files.
Share bugs, ideas, or general feedback.
Own this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge.
Sign in to claimOwn this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge.
Sign in to claimBased on adoption, maintenance, documentation, and repository signals. Not a security audit or endorsement.
Orchestrate multi-agent teams for parallel code review, hypothesis-driven debugging, and coordinated feature development using Claude Code's Agent Teams
Multi-agent orchestration for Claude Code. 12 specialized agents working in parallel — planning, building, reviewing, debugging. Plus a Hub for always-alive multi-project sessions controllable from Telegram or Slack.
Multi-agent team bootstrapper and runtime orchestrator — analyzes projects and generates tailored Claude Cadre teams via AI debate (init), then coordinates live agent teams for parallel review, debate, planning, building, and research (orchestrate)
Multi-agent team orchestration for parallel task execution, research, and implementation
High-intelligence Claude Code copilot with deep code reasoning, evidence-driven planning, orchestration-first execution, model routing, context budgeting, CI/CD integration, enterprise security, plugin development, prompt engineering, performance profiling, agent teams, channels (event-driven autonomy with CI webhook, mobile approval relay, Discord bridge, and fakechat dev profile), interactive tutorials, LSP integration, security-hardened hook script library, MCP Prompts coverage, common workflow packs, runtime selection guide, computer-use patterns, checkpointing, scheduled-task blueprints, repo bootstrap scanner, hook policy engine (8 installable packs), layered memory deployment, role-based subagent packs (implementer, debugger, migration-lead, dependency-auditor, release-coordinator), 5 agent-team topology kits, autonomy operating mode (4 profiles + 3 gates), and a queryable 15-tool MCP documentation server with autonomy advisor.
Intelligently compose and deploy Claude Code Agent Teams. Auto-selects optimal team composition from 24+ agents and 83+ skills across 5 scopes, generates task dependency graphs, and orchestrates multi-agent workflows with a single command.
https://github.com/user-attachments/assets/a66620cb-4a81-4525-95f2-1f1f22765288
Agent Teams reinvented for collaborative coding supporting Claude Code and Codex, with full code transparency.
CAFleet works with two coding agents: claude (Claude Code) and codex (OpenAI Codex CLI). Install the plugin in whichever one you use — the broker CLI is shared.
Run the following commands on your terminal:
/plugin marketplace add himkt/cafleet
/plugin install cafleet@cafleet
[!IMPORTANT]
Make sure whether the skills are correctly installed. You can see available skills by running
/skillson Claude Code prompt.Typically, the config entries like following would be written in
~/.claude/settings.json:{ "permissions": { "allow": [ "Bash(cafleet *)", "Skill(cafleet:agent-team-monitoring)", "Skill(cafleet:agent-team-supervision)", "Skill(cafleet:base-dir)", "Skill(cafleet:cafleet)", "Skill(cafleet:create-figure)", "Skill(cafleet:design-doc)", "Skill(cafleet:design-doc-create)", "Skill(cafleet:design-doc-execute)", "Skill(cafleet:design-doc-interview)", "Skill(cafleet:my-slidev)", "Skill(cafleet:research-presentation)", "Skill(cafleet:research-report)" ], "ask": [ "Bash(cafleet * member exec *)" ] }, "enabledPlugins": { "cafleet@cafleet": true }, "extraKnownMarketplaces": { "cafleet": { "source": { "source": "github", "path": "himkt/cafleet" } } } }
Run the following command on your terminal
codex plugin marketplace add himkt/cafleet
[!IMPORTANT]
Make sure whether the skills are correctly installed. You can see available skills by running
/skillson Codex prompt.Typically, the config entries like following would be written in
~/.codex/config.toml:[marketplaces.cafleet] last_updated = "2026-05-16T11:56:51Z" last_revision = "03a8caa66c8a7981345d74fe3aec9a6e498792a1" source_type = "git" source = "https://github.com/himkt/cafleet.git" [plugins."cafleet@cafleet"] enabled = true
[!IMPORTANT] Codex members need the cafleet DB directory to be writable from inside the codex sandbox. Add it to
sandbox_workspace_write.writable_rootsin anyconfig.tomlcodex reads (e.g.~/.codex/config.toml):[sandbox_workspace_write] writable_roots = ["/home/<you>/.local/share/cafleet"]Use the absolute path matching
CAFLEET_DATABASE_URLor the default XDG location.
[!TIP]
Recommended Codex rules for
cafleetcommands — drop into your codex rules file (e.g.~/.codex/rules/cafleet.rules) to auto-approve the broker CLI while keepingcafleet member exec(which runs arbitrary shell on a member pane) gated behind a prompt:prefix_rule( pattern = ["cafleet"], decision = "allow", justification = "All cafleet subcommands are allowed by default", ) prefix_rule( pattern = ["cafleet", "member", "exec"], decision = "prompt", justification = "cafleet member exec runs arbitrary commands on a member; require approval", )
For codex CLI version pin and operational specifics, see docs/codex-members.md.
uv tool install cafleet # or: pip install cafleet
cafleet db init # apply schema migrations (idempotent; rerun after upgrades)
The default database is ~/.local/share/cafleet/registry.db. Override with CAFLEET_DATABASE_URL (use an absolute path — SQLAlchemy does not expand ~ in SQLite URLs).
Provide the following prompt to Claude Code or Codex to see how it works.
I want to see how cafleet works.
Please create a fresh team with two teammates using cafleet and let them ping-pong each other.
After the demonstration, please shutdown the team.
CAFleet provides the builtin skills for Spec Driven Development (SDD). We're using CAFleet to develop CAFleet!
Claude Code:
/cafleet:design-doc-create I want to create a simple TUI calculator
Codex:
$cafleet:design-doc-create I want to create a simple TUI calculator
You can see the existing design docs on design-docs/, which are actually created by the skills.