delegation plugins

by spacegrowth

Delegate work packets from a lead Claude Code session to executor sessions in their own iTerm tabs, with session reuse instead of always spawning fresh.

by Abhirup Banerjee

Delivery discipline for AI coding agents. Writes the five-layer Stackwich policy (Prompt / Context / Harness / Loop / Graph) into CLAUDE.md and scaffolds advisor, executor and verifier subagents.

by VlDubr

Bidirectional bridge between Claude Code and Codex (ChatGPT): code review, adversarial review, task delegation, background jobs, models consulting each other, and image generation via Codex's built-in tool. Runs on your existing ChatGPT subscription, no API keys.

by AER Engineering

Adaptive repository-aware engineering control plane for coding, debugging, research, POCs, reviews and safe changes, with capability-aware providers, durable memory, procedural skills, bounded delegation, secure tools, verification and evidence-backed learning.

by Arthur Carroll

Delegates long-running coding and research tasks to the OpenCode and Antigravity (agy) CLIs, and to the Gemini web app, as unattended MCP subagents. Dispatch returns a run id immediately and records the run on disk, so a delegate stays reachable - checkable, tailable, cancellable - across dropped connections, session ends and server restarts. Fails fast on provider quota walls instead of blocking for the full hour, kills hung delegates on an idle timer, and reports whatever the delegate wrote to .agent-runs/ on every exit path. The Gemini wrapper drives a dedicated Chrome profile to reach live Google search, Canvas, Gems and conversation history, which no Gemini API exposes.

by TamBN3

Bộ điều phối subagent cho Claude Code với phê duyệt kế hoạch theo prompt, biên nhận kiểm chứng có cấu trúc và kiểm tra glossary khớp chính xác.

by Andre Paat

Multi-model delegation for browser testing — Kimi K2.5 (via opencode) for execution, Opus for analysis

by Bymax One

Engineering Project Manager for multi-agent development. The /bymax-pm:pm command turns the current session into a PM/TPM that coordinates INDEPENDENT Claude Code peer sessions: discovers agents via ListAgents, delegates structured task contracts via SendMessage, tracks a deterministic task lifecycle (assignment → acknowledgement → evidence → independent review → PM verification), detects stuck agents through one-shot idle notices instead of polling, resolves blockers between agents before escalating, and keeps all project truth in a persistent, git-friendly .claude/pm/ workspace that survives session restarts. Evidence-driven: a worker saying done is never enough — DONE requires diffs, tests, CI and PM verification. The PM coordinates; it does not implement.

by abgonzalez93

Keeps whole-file reads and boilerplate writing out of the expensive model's context by delegating them to a cheap one-shot worker, only in the projects you plug in.

by claude-skills

Routes work from an expensive lead model to tool-scoped worker agents -- explorer and researcher on Haiku, editor and test-runner on Sonnet, refuter and investigator on Opus -- and makes the lead re-verify what they report instead of trusting it. Adds fabflows:build, a deterministic loop in which an Opus builder implements a spec and a fresh reviewer returns ACCEPT or REWORK. Ships an ACTIVE guard hook that blocks package installs, commits and pushes to a default branch, destructive shell commands, and reads or writes of credential files; the guard fails open, so a bug degrades to no guard rather than a blocked session.

by Dmitry Fomin

Route a feature through four stages — spec plus acceptance checklist, an outside opinion, implementation in another agentic harness (Grok Build or DeepSeek), and an independent verification — while the main context only orchestrates and never writes the code itself.

by dd106-lang

Keep the expensive lead model on planning and review; route routine work to pinned cheaper workers (Haiku, Sonnet, Opus) and audit which models actually ran.

by jellologic

Offload bulk coding work to Muse Code instances in isolated git worktrees, each supervised by an Opus agent that reads the patch, runs the acceptance check and re-prompts until the work is right.

by Kourosh Mobl

Pool several AI subscriptions inside Claude Code by delegating to vendor CLIs as subprocesses. Ships field-tested adapters for GLM (Z.AI), Codex (ChatGPT), Copilot, Antigravity (Gemini), Ollama (local) OpenRouter (metered, many vendors behind one key) and a second Claude subscription, a multi-model consultation panel, task delegation into throwaway worktrees you review as a diff, and a probe-driven generator that builds and verifies an adapter for any provider you already have.

by HernandoR

Model-tiered Claude Code subagents: haiku for mechanical execution, opus for judgment-heavy development, fable for read-only review

by Sebastian Rädle

Werkzeuge mit ausschließlich expliziter Auslösung: /tools:ask und /tools:execute (die jeweils andere CLI non-interaktiv fragen bzw. beauftragen), /tools:unslop (KI-Muster aus Text entfernen), /tools:grimm (Sachverhalte ins Verwaltungs- und Anordnungsdeutsch überführen), /tools:orchestrator[-review|-cross] (Rollen-Prompt: steuern statt arbeiten, in drei Stufen bis zur anderen CLI als Arbeiter). Nachfolger der Plugins ask, unslop und grimm.