subagent-swarm-mcp
Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type
@followed by the MCP server name and your instructions, e.g., "@subagent-swarm-mcp offload a quick readonly review of src/server.mjs to a subagent"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
subagent-swarm-mcp
简体中文 | English
学习研究与合理使用声明: 本项目为非官方开源社区项目,仅用于个人学习 MCP(Model Context Protocol)协议编排、多 CLI 工具集成与本地交互式开发工作流研究。本项目不修改任何官方客户端代码,也不包含任何绕过鉴权、限流或计费的机制。使用前请务必遵守 智谱 AI(Zhipu AI / Z.ai / ZCode)、腾讯(CodeBuddy) 及 Qoder 各自的《用户协议》《服务条款》与合理使用政策,严禁用于商业转售、账号共享或高频自动化刷量,请克制调用频率与并发,切勿给智谱及其他平台服务端造成过大负担;大规模批量任务请使用平台官方正式按量 API / Batch API 或本地部署模型。详情请务必阅读文末《安全、合规与免责声明》。
一个运行在 Linux(Node.js >= 24)上的多 CLI 子代理调度 MCP Server 与命令行工具集(前身为 glm-zcode-mcp),用于探索和学习如何让 Claude Code、Codex、Antigravity 等 Agent 在本机交互式开发场景下,按任务复杂度与权限需求将轻量辅助任务低开销地分发到多个已合法登录的编程工具(目前已内置 CodeBuddy CLI、Qoder CLI 与官方 ZCode 桌面端,架构通过可插拔注册表预留扩展接口)。
架构与后端支持
┌──> CodeBuddy CLI (`cb_*`, 独立脱附 worker) ─> codebuddy -p --output-format json
agent ──MCP/stdio──> subagent-swarm-mcp (`swarm_*` / `zcode_*` / `bin/swarm.mjs`)
├──> Qoder CLI (`qd_*`, 独立脱附 worker) ─────> qodercli -p -o json
└──> ZCode (`sess_*`, CDP 127.0.0.1:9222) ──> 官方 ZCode 桌面窗口 + 只读 SQLite/JSONLCodeBuddy & Qoder CLI 后端(
codebuddy、qoder): 通过独立进程组 worker(cb_<hex>/qd_<uuid>)与 stdin 管道安全调用对应官方 CLI,原生提供 3 级权限/工具隔离模式(lite零工具超低开销、readonly只读检视且 CodeBuddy 内含ViewImage多模态看图、workspace全量工作区权限)、支持waitSec单次内联拿结果、-r多轮续聊、-pgid整棵进程树终止、孤儿进程 fail-close 回收与实时 JSONL 工具状态观测。ZCode 桌面端后端(
zcode): 仅通过 Electron 本地回环 DevTools 端口(127.0.0.1:9222)与跨进程mkdir文件锁配合官方 ZCode 桌面窗口操作(不修改官方客户端文件,不提取或重放账号凭据,也不直接调用外部 ZCode CLI),严格校验计费路由白名单(ZCODE_MCP_PROVIDERS),只读读取本地 SQLite 与 JSONL 日志。原有 29 个zcode_*、hub_*与codebuddy_usage工具保持兼容。智能
auto路由: 按调用约束(lite/readonly/ephemeral/ 自定义cwd/ 专属模型 / 专属计费路由)、后端健康度与当前空闲并发槽位自动在可用 CLI 与 ZCode 间分流。
Related MCP server: Pair
快速上手
前置条件: Linux(
x86_64/arm64)、Node.js >= 24,并至少就绪其中一个您本人合法登录的后端:CodeBuddy CLI: 已安装并登录
codebuddy(可执行codebuddy --version)。Qoder CLI: 已安装并登录
qodercli(可执行qodercli status)。ZCode 桌面端: 官方 ZCode(已实测
3.16.1/3.18.2)以scripts/start-zcode.sh(或scripts/zcode-headless.sh)启动,已登录个人 Coding Plan,且左侧栏有指向~/projects/subagents的subagent项目。
安装并注册到 MCP 客户端:
git clone https://github.com/wst-ki/subagent-swarm-mcp.git && cd subagent-swarm-mcp && npm install # Claude Code claude mcp add subagent-swarm-mcp -s user -- node "$PWD/src/server.mjs" # Antigravity (agy) agy mcp add subagent-swarm-mcp node "$PWD/src/server.mjs" # Codex CLI codex mcp add subagent-swarm-mcp -- node "$PWD/src/server.mjs"或在 Codex(
~/.codex/config.toml)中手动加入:[mcp_servers.subagent-swarm-mcp] command = "node" args = ["/absolute/path/to/src/server.mjs"] tool_timeout_sec = 120
核心能力一览
统一子代理池(
swarm_backends、swarm_run、swarm_wait、swarm_status、swarm_result、swarm_followup、swarm_cancel、swarm_list、swarm_quota): 跨auto、codebuddy、qoder、zcode统一调度;对高响应 CLI 模型(如hy4flash-free、deepseek-v4.1-flash、DeepSeek-Flash)可传入waitSec: 30..55单次直接拿到结果;无 MCP 环境可使用bin/swarm.mjs。高稳健结构化输出与小批量任务编排(
swarm_run_structured、swarm_structured_result、swarm_batch_run、swarm_batch_status): 支持受限 JSON Schema 与domainRule校验、单轮纯 JSON 协议(绕开 CodeBuddy--json-schema与--tools ""碰撞)、JSONL 断点续跑、微批打包、部分失败单独重试、累计usage.credit预算熔断(--max-credit/maxCredit)、多任务动态槽位礼让(--reserve-slots/reserveSlots)、跨进程槽位忙等待退避(capacityWaitMs)、机器可读阻塞分类(blocked.json含retryable与recommendedAction)、实时进度心跳(progress.json)、单项超时自动取消防槽位泄漏、额度熔断恢复零漏篇续跑、断点部分结果抢救导出(swarm batch-export/zcode batch-export)、原子文件落盘(--summary/--export)、精简输出(--compact/includeItems: false)与防>64 KiB管道截断刷新。深度消费与成本观测(
codebuddy_usage、bin/codebuddy-usage.mjs、swarm_quota): 合并统计 CodeBuddy 本地 CLI 会话记录(~/.codebuddy/projects/)与 swarm 脱附任务状态(<stateDir>/cb-runs/,完整覆盖--no-session-persistence/ephemeral临时任务且自动去重持久化会话),输出临期额度包告警(expiringSoon)、Token 与缓存命中率/思考 Token 明细、按模型/权限档位(lite/readonly/workspace)单次成本核算、日消耗趋势及余量续航预测(runway);同步保留旧版 ZCode 与 Dispatch Hub 能力(zcode_*、zcode_session_*、hub_*、bin/zcode.mjs、bin/zcode-watch.mjs、bin/dispatch.mjs、hub-kit/)。
全部 42 个工具参数、CLI 参数、环境变量与技术限制详见 docs/REFERENCE.md、docs/HUB-PLAN.md 与 docs/SUPERVISOR.md。
常用配置
环境变量 | 默认值 | 说明 |
| 自动探测 | 显式指定 |
|
| CodeBuddy 默认模型 |
|
| CodeBuddy 并行任务上限 |
|
| CLI 子代理默认工作目录 |
| 自动探测 | 显式指定 |
|
| Qoder 默认模型 |
|
| Qoder 并行任务上限 |
|
| ZCode 本地 DevTools 调试端口( |
|
| ZCode 计费路由白名单 |
|
| ZCode 并行任务上限 |
全部环境变量(ZCODE_MCP_*、CODEBUDDY_*、QODER_*、HUB_*、ZCODE_MCP_LOCAL_MODULE)详见 docs/REFERENCE.md。
开发与测试
npm test # 运行离线单元测试(无需实机 CLI 或 ZCode)
ZCODE_MCP_LOCAL_MODULE=none npm test # 验证无私有扩展模块时的纯净模式
node e2e/client.mjs swarm_backends # 通过真实 MCP stdio 自检各后端连通性安全、合规与免责声明
本地执行权限边界: 在 workspace 模式以及所有 ZCode 任务(会自动切换为 完全访问 / YOLO)下,子代理拥有当前操作系统用户的完整本地权限。纯文本推理、抽取与分类任务强烈建议优先使用 CodeBuddy / Qoder 的 profile: "lite"(零工具)或 "readonly";严禁将本地调试端口(127.0.0.1:9222)暴露至公网或不可信局域网,也不要向具有工作区写权限的子代理传入未经审查的外部不可信输入。
仅限个人学习与技术研究: 本项目为开源技术学习与实验性参考实现,旨在演示 MCP Server 开发、多 CLI 抽象适配、结构化输出校验与本地任务编排架构,不以任何商业盈利为目的,亦不鼓励或支持任何形式的违规自动化滥用。
严格遵循各平台服务协议与合理使用原则:
使用本项目对接 智谱 AI(Zhipu AI / Z.ai / ZCode / GLM Coding Plan)、腾讯(CodeBuddy)、Qoder 等第三方服务时,使用者必须自行阅读并严格遵守各平台现行的《用户协议》《服务条款》《使用须知》及配额/速率限制规定(此类个人订阅套餐通常仅限自然人个人交互式软件开发辅助,明确禁止多人共享账号、API 转售/聚合代理、非交互式大规模爬取或恶意刷量)。
请文明、克制使用,切勿给智谱及其他平台服务器造成过大负担: 请将并发数与任务规模控制在个人日常交互开发合理范围内,避免长时间 7×24 高频满载打满配额窗口;如确有大规模、长时间或生产级批量数据处理需求,请务必使用各平台官方正式提供的按量计费开放 API / Batch API 或本地自部署开源模型。
技术合规边界(零破解、零篡改、零窃密): 本项目不修改、不反编译、不注入任何官方客户端程序或二进制文件,不提供任何绕过登录认证、限流保护、计费体系或提取/重放第三方鉴权凭据的功能,也不向作者或任何第三方收集、上传用户数据与账号密钥。
无关联声明与商标归属: 本项目为独立非官方社区项目,与智谱 AI(Zhipu AI / Z.ai)、腾讯(Tencent)、Qoder 或其他任何提及的软件与平台均无任何关联、合作、授权、背书或赞助关系;文中涉及的产品名称与商标均归其各自合法权利人所有。
风险自担与责任豁免: 本软件依据 MIT License 按“现状(AS-IS)”提供,不附带任何明示或默示担保。因下载、部署、修改、配置或不当使用本项目所导致的任何直接或间接后果(包括但不限于平台限流、风控拦截、套餐权益暂停、账号封禁、额外计费损失、本地文件改动或数据损毁等),均由使用者本人自行承担全部法律与经济责任,项目作者及贡献者在法律允许的最大范围内不承担任何责任。
权利人沟通与配合: 如智谱 AI 或其他相关平台权利人认为本项目的任何部分存在合规、条款冲突或其他疑虑,请随时通过仓库 Issue 联系,作者将在收到通知后第一时间积极配合核查、调整或下架相关内容。
第三方开源代码归属声明见 THIRD_PARTY_NOTICES.md。
Available Tools
42 toolscodebuddy_usageCodeBuddy credit & cost analyticsARead-only
CodeBuddy/WorkBuddy account credit (read-only billing API, cached 10 min, with expiring-soon package alerts) + deduplicated local consumption across both CLI JSONL transcripts and swarm cb-runs state files (including ephemeral runs): tokens & cache-hit rate, per-model/per-profile unit economics, daily breakdown, top tasks, and burn-rate runway forecast.
Inputs: refresh, sinceDays, expiringSoonDays.
hub_answerAnswer a hub task's questionA
Reply to a task in need-info; it goes back to the same worker (a ZCode worker continues its session with the answer).
Inputs: id (required), text (required).
hub_assignOverride a hub task's assignmentBIdempotent
Assign a queued (or assigned) task to a specific worker yourself, bypassing the scheduler. See hub_board / the workers roster for ids.
Inputs: id (required), worker (required), reason.
hub_boardList hub tasksARead-only
All tasks in the dispatch hub with state, assignee and latest note. States: queued, assigned, running, done, need-info, failed, accepted, rejected, cancelled.
Inputs: no named parameters.
hub_cancelCancel a hub taskBDestructiveIdempotent
Cancel a task that is not finished. Terminal: accepted, cancelled (and failed/rejected until requeued).
Inputs: id (required).
hub_logHub task timelineARead-only
One merged, time-sorted line per event on a task: state changes (the hub's commit subjects), worker PROGRESS lines and INBOX messages. A cheap way to see how a task is going without hub_show's full payload; for unattended watching prefer dispatch watch (README).
Inputs: id (required).
hub_postPost a task to the dispatch hubA
Queue a task for the dispatch hub's workers (ZCode / CLI agents on the hub host, manual workers on remote machines). Returns once the task is queued; a daemon assigns and runs it, and hub_wait wakes you when it needs attention. The task body must be self-contained (the worker cannot see this conversation). Code tasks need repo (absolute path on the hub host) and base (commit sha); the worker's changes come back as a patch that is replayed and checked by the hub's acceptance step.
Inputs: task (required), title (required), base, kind, +8 more.
hub_requeueRequeue a hub taskA
Put a failed / rejected / cancelled task back into the queue with its assignee cleared.
Inputs: id (required).
hub_showShow one hub taskBRead-only
Full task record: spec, note, the instructions given to the worker, the master answer, and (once done) the worker's report.
Inputs: id (required).
hub_tellMessage a hub task's workerA
Append a short message (collapsed to one line, max 300 chars) to the task's inbox. The worker sees INBOX lines on its next hub say/inbox/next and is expected to follow them. Use for nudges and clarifications that do not need a state change (unlike hub_answer, which replies to need-info).
Inputs: id (required), text (required).
hub_waitWait for hub eventsARead-only
Block until the hub needs the master's attention (a task entered need-info / failed / rejected, or the hub went idle after a change) or timeoutSec elapses (default 55, max 55). Returns { timedOut, lines }; if timedOut, just call hub_wait again.
Inputs: until, cursor, timeoutSec.
swarm_backendsList available swarm backends and live healthARead-only
Inspect registered subagent CLI/UI backends (codebuddy, qoder, zcode), their availability, supported models, profiles, and active/max concurrency.
Inputs: no named parameters.
swarm_batch_runRun or resume a checkpointed structured swarm batchCDestructiveIdempotent
Execute or resume a checkpointed structured batch (with microbatching, partial recovery, and circuit breakers) across 'auto' (default), 'codebuddy', or 'zcode'.
Inputs: task (required), items (required), schema (required), batchId (required), +18 more.
swarm_batch_statusRead checkpointed swarm batch statusBRead-only
Read the current manifest, item statuses, aggregated token/credit usage, and any blocker for a batchId.
Inputs: batchId (required), includeItems.
swarm_cancelCancel a running swarm subagent taskCIdempotent
Terminate a running cb_* process group or click Stop on a running sess_* ZCode task.
Inputs: taskId (required).
swarm_followupSend a follow-up prompt to a finished swarm taskBDestructive
Continue a finished, non-ephemeral cb_* or sess_* task in the same conversation session.
Inputs: prompt (required), taskId (required), model, fallbackModel.
swarm_listList recent swarm tasks across backendsBRead-only
List recent tasks from CodeBuddy, Qoder, ZCode, or all merged newest-first.
Inputs: limit, backend.
swarm_quotaRead quota across swarm backendsBRead-only
Read available quota/usage information from CodeBuddy, Qoder, ZCode UI, or all (all).
Inputs: backend, refresh, sinceDays.
swarm_resultGet swarm subagent task resultARead-only
Return the final result text, model info, and token/credit usage for a finished cb_* or sess_* task.
Inputs: taskId (required).
swarm_runRun a subagent task on an available CLI backendADestructive
Delegate a task to the selected CLI backend ('auto', 'codebuddy', 'qoder', or 'zcode'). On CLI backends (codebuddy, qoder), profile defaults to 'lite' (zero tools, minimal token/credit cost; great for fast pure-reasoning/extraction), readonly enables Read/ViewImage/Glob/Grep/WebSearch/WebFetch, and workspace grants full workspace access in cwd. Set waitSec (e.g. 30-55) for single-call inline completion on fast CLI tasks, or omit / 0 to return taskId immediately.
Inputs: prompt (required), cwd, model, backend, +5 more.
swarm_run_structuredRun a structured JSON-contract task on a swarm backendBDestructive
Execute a bounded JSON Schema + optional domainRule task on 'auto' (default), 'codebuddy', or 'zcode'. CLI backends default to zero-tool 'lite' profile; set waitSec > 0 to wait and validate inline.
Inputs: task (required), input (required), schema (required), cwd, +8 more.
swarm_statusGet swarm subagent task statusARead-only
Non-blocking status check for any cb_* or sess_* task, reconciling dead worker processes automatically.
Inputs: taskId (required).
swarm_structured_resultValidate a structured swarm task resultARead-only
Recover the stored JSON Schema + domainRule contract for a cb_* or sess_* task and validate its output.
Inputs: taskId (required).
swarm_waitWait for any swarm subagent taskARead-only
Block until a cb_* or sess_* task reaches a terminal state or timeoutSec elapses (default 45, max 55). Returns status and final result when complete.
Inputs: taskId (required), timeoutSec.
zcode_batch_runRun or resume a checkpointed structured ZCode batchCDestructiveIdempotent
Execute or resume a low-CPU structured batch with deterministic item keys, append-only JSONL checkpoints (manifest.jsonl, results.jsonl, summary.json, blocked.json), bounded retry (0 or 1), and circuit breaking on quota/rate-limit, login blockers, or provider route mismatch.
Inputs: task (required), items (required), schema (required), batchId (required), +13 more.
zcode_batch_statusRead checkpointed ZCode batch statusBRead-only
Read the current checkpointed manifest, item statuses, aggregated token usage, and any circuit-breaker blocker for a batchId.
Inputs: batchId (required), includeItems.
zcode_cancelCancel a running ZCode taskAIdempotent
Stop a running task (clicks ZCode's stop button). Work already done on disk is not rolled back.
Inputs: taskId (required).
zcode_followupSend a follow-up to a ZCode taskADestructive
Continue a FINISHED task with another message in the same ZCode conversation (it keeps its earlier context). Returns immediately; use zcode_wait afterwards. The task must not be running, and must still be visible in ZCode's sidebar (not archived).
Inputs: prompt (required), taskId (required).
zcode_healthCheck the ZCode connectionARead-only
Verify ZCode is running with its debugging port, the subagent project exists, and report how many tasks are running.
Inputs: no named parameters.
zcode_listList recent ZCode tasksCRead-only
Recent (non-archived) tasks in the subagent project with their status.
Inputs: limit.
zcode_progressGemini progress digest for a ZCode taskARead-only
Cheap Gemini Flash digest (via the agy CLI) of what the task's subagent has done so far: summary, done/doing/next, an estimated progressPct against the task prompt, onTrack, concerns and needsAttention. Works while the task runs and after it finishes (summary of the whole run). Digests are cached per task for ZCODE_MCP_SUMMARY_MIN_SEC seconds (default 300) and only recomputed when new activity exists; set refresh to bypass the cache. For unattended watching prefer bin/zcode-watch.mjs (see README) over polling this.
Inputs: taskId (required), refresh.
zcode_quotaRead displayed ZCode quotaARead-onlyIdempotent
Read the remaining quota percentages displayed by the official ZCode UI on demand. May be cached; account and provider cannot be verified, reset times are display text, and missing windows mean unknown. Temporarily opens and closes the context popover; never presses reset or reads credentials.
Inputs: no named parameters.
zcode_resultGet ZCode task resultARead-only
Final answer text of a finished task (plus token usage). For a task still running this returns its status without a result.
Inputs: taskId (required).
zcode_runStart a ZCode subagent taskADestructive
Delegate a task to a GLM subagent running in the official ZCode desktop app (uses the user's GLM Coding Plan quota). Returns a taskId immediately; the task runs in the background. Use zcode_wait to block until it finishes, or zcode_status to poll. The subagent runs with FULL ACCESS (shell, file read/write anywhere the OS user can) in the project folder /root/projects/subagents. It cannot see this conversation, so put ALL needed context in the prompt: goal, absolute file paths, constraints, and the exact output format you want back. Prefer small, well-scoped tasks. Several tasks may run in parallel (limit 3). Models: GLM-5.3 | GLM-5.3-Flash (default GLM-5.3; Flash is faster and cheaper for simple work). provider picks the billing route (the model menu lists the same names under both): account%3Abigmodel-individual-coding-plan = BigModel 个人 (default), account%3Abigmodel-start-plan = Start Plan 免费.
Inputs: prompt (required), model, provider.
zcode_run_structuredStart a structured ZCode taskADestructive
Start a normal ZCode session with a bounded JSON schema contract and optional domain semantic rule preset (e.g. section-outline-v1, or a declarative rule object). This uses subscription quota and FULL ACCESS like zcode_run. The schema and domainRule are stored in the session prompt; use zcode_structured_result after completion.
Inputs: task (required), input (required), schema (required), model, +2 more.
zcode_session_eventsPoll ZCode session lifecycle eventsARead-only
Read a bounded page of existing ZCode JSONL lifecycle events for a session. Pass the returned cursor on subsequent calls. This is not the v4 stream: assistant text and tool arguments/results are unavailable. The first call starts at the newest log by default, or the oldest retained log with start=oldest; gaps report incomplete history, rotation, truncation or unreadable records.
Inputs: taskId (required), limit, start, cursor.
zcode_session_model_ioRead an optional ZCode model IO recordARead-only
Read one retained model IO JSONL record for a session. Recording may be absent or incomplete. Returns metadata and tool names by default; includeContent also returns assistant text and tool arguments. Never returns request headers or full prompt; not a per-token event stream. Records over 1 MiB are reported and not skipped.
Inputs: taskId (required), cursor, includeContent.
zcode_session_part_contentRead a session part in bounded chunksARead-only
Read UTF-8 JSON bytes of an assistant text or tool snapshot in base64 chunks. Pass version as expected on subsequent pages; updates normally invalidate the read, but same-millisecond same-size changes may be missed. Content may contain task text, tool arguments and output.
Inputs: partId (required), taskId (required), limit, offset, +1 more.
zcode_session_partsRead a bounded ZCode session snapshotARead-only
Read a page of assistant text and tool part snapshots from ZCode's SQLite database. This is not a live event stream: parts can change and pagination cannot detect changes to earlier rows. Content is omitted by default; opting in may expose task text, tool arguments and results. Parts above 8192 bytes are omitted, and the cursor sorts by ID, not chronological order.
Inputs: taskId (required), after, limit, includeContent.
zcode_statusGet ZCode task statusARead-only
Non-blocking status of a task: running | completed | failed | cancelled | starting. While running it includes elapsedSec, the tool currently executing, tool call count and idleSec (seconds since the last activity; stalled: true if suspiciously long). Does not include the answer — use zcode_result or zcode_wait for that.
Inputs: taskId (required).
zcode_structured_resultValidate a structured ZCode resultARead-only
Recover the original schema and optional domainRule/input from the session and validate the final JSON answer. Invalid, unfinished, failed and cancelled tasks never count as valid.
Inputs: taskId (required), domainRule.
zcode_waitWait for a ZCode taskARead-only
Block until the task finishes or timeoutSec elapses (default 45, max 55 so it stays under typical 60s tool-call timeouts), then return its status; when finished the result text is included. If it returns status 'running', just call zcode_wait again. Prefer this over tight polling.
Inputs: taskId (required), timeoutSec.
Tool Schema Changelog
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
TDQS
Scored across 42 tools
Two large families duplicate each other almost one-for-one: zcode_run/swarm_run, zcode_wait/swarm_wait, zcode_result/swarm_result, zcode_status/swarm_status, zcode_list/swarm_list, zcode_cancel/swarm_cancel, zcode_followup/swarm_followup, zcode_quota/swarm_quota, plus zcode_run_structured vs swarm_run_structured and zcode_structured_result vs swarm_structured_result. There are also three blocking 'wait' tools (hub_wait, zcode_wait, swarm_wait) and multiple progress-inspection paths (hub_log, hub_show, zcode_status, zcode_progress) whose boundaries an agent must infer from dense prose. Descriptions do try to draw lines (e.g. swarm_run subsumes zcode as a backend), but the overlap is severe enough to cause frequent misselection.
Within each prefix the verb_noun pattern holds (hub_post/hub_board/hub_show, zcode_run/zcode_wait/zcode_result, swarm_run/swarm_status), and prefixes are used consistently as namespace markers. However, the namespaces themselves are not parallel: the same operations exist under both zcode_* and swarm_*, and one tool (codebuddy_usage) is named after a backend while the equivalent exists as swarm_quota, so the scheme doesn't tell an agent which prefix to reach for.
42 tools for what is essentially task delegation, monitoring, structured/batch runs, a dispatch hub, and session log reading. Roughly fifteen of those are near-aliases of others, so the count is inflated well beyond the real surface area. The hub_* group is reasonably scoped on its own, but the overall set is too large for the domain.
The lifecycle is well covered: submit (swarm_run/zcode_run/hub_post), observe (status/board/log/progress), block (wait), collect (result/structured_result), continue (followup/hub_answer/hub_tell), stop (cancel), recover (requeue/batch status), and cost (quota/usage) all exist, plus session-log forensics. Gaps are minor, e.g. there is no operation to archive/delete a hub task or to enumerate the worker roster directly (hub_assign only points at it).
Maintenance
Related MCP Connectors
Multiplayer for the coding agents your team already uses: shared tasks, decisions, skills and tools.
Command center for developers coordinating AI coding agents across projects.
Backend platform for AI agents: one agentstack.execute tool runs batched steps over +700 actions (hosting, storage, auth, payments, CRM, app builder).
Hosted runtime for persistent agent teams, durable workflows, memory, schedules, and goals.
Related MCP Servers
- AlicenseBqualityBmaintenanceEnables Claude Code and Claude Desktop to delegate bounded, disposable tasks to local or cheap cloud models, with model scoring, guardrails, and cost-saved tracking.53MIT
- AlicenseNot gradedqualityCmaintenanceEnables users to delegate coding tasks in a chosen project to the official Codex and Claude CLI agents from within an existing chat, then convene a council of configured models to gather independent opinions and review the resulting files and test runs. It also manages multi-provider API connections, model catalogs, and per-task token, time, and cost limits.MIT
- AlicenseNot gradedqualityAmaintenanceEnables hosts like Claude Code, Codex, Antigravity and Grok to dispatch tasks to DeepSeek Harness agents that appear as native subagents with live progress and token stats. It provides tier and effort selection, escalation and dispatch guardrails, a jobs board, in-host or standalone worker sessions, plus a multimodal bridge that lends text-only models vision and image generation.354 npmMIT
- AlicenseNot gradedqualityAmaintenanceEnables AI coding providers to run structured multi-agent workflows with tracked tasks, document specifications, handoffs, health gates, and server-owned evidence across providers like Claude Code, Codex CLI, Cursor, Grok Build, and OpenCode.29 npmMIT