Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx skills add wu736139669/agent-taskgraph-protocol --skill agent-taskgraphgit clone --depth 1 https://github.com/wu736139669/agent-taskgraph-protocolWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/wu736139669/agent-taskgraph-protocol/agent-taskgraph)<a href="https://agentmods.dev/skills/wu736139669/agent-taskgraph-protocol/agent-taskgraph"><img src="https://agentmods.dev/badge/skills/wu736139669/agent-taskgraph-protocol/agent-taskgraph/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/wu736139669/agent-taskgraph-protocol/agent-taskgraph"><img src="https://agentmods.dev/badge/skills/wu736139669/agent-taskgraph-protocol/agent-taskgraph.svg" alt="Reviewed on agentmods" width="80" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00097 | $0.01739 |
| Opus 5.5 | $0.00039 | $0.00696 |
| Sonnet 5.5 | $0.00019 | $0.00348 |
| Haiku 4.5 | $0.00010 | $0.00174 |
Grade A, and why
agent-taskgraph scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 1mo ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
This is a copy
100% identical to agent-taskgraph — 0 lines differ, which has more behind it and is treated as the original. This page carries a canonical link to it rather than competing with it.
How it starts
The opening of the file, as written. The whole thing — 97 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Agent TaskGraph
使用当前宿主的原生 Agent 能力完成工作。Team 是逻辑协作模型，不等同于“启动多个进程”，也不需要外部控制面或自建队列。
1. 选择最小有效模式
先读适用的 AGENTS.md / CLAUDE.md、相关代码和验证命令。
| 模式 | 选择条件 | 形态 |
|---|---|---|
| Solo | 小修复、单模块、顺序工作，拆分收益低 | 当前会话直接完成 |
| Delegation | 多个独立支线只需分别交给 Lead | Lead + 原生 Members |
| Team | 有共同目标、角色协作、任务依赖、共同决策或跨模块集成 | Team Charter + Task Graph + 协作协议 + 生命周期 |
用户明确要求“组团队”时优先使用 Team。只有原生 Agent 不可用、任务无法拆出独立职责，或写入无法安全隔离时才降级，并说明原因。
默认包含 Lead 在内共 2-4 个有真实职责的 active Members。不要为组织结构完整而创建空闲 Agent。调用本 Skill 已授权合理的内部拆分和原生 Agent 创建；只有代码无法回答的产品决策、权限扩大或外部副作用需要提问。用户明确要求先看分工时才等待确认。
2. 建立 Team
Team 必须具备：
- 一个可观察的共同目标、非目标和完成条件。
- Lead、Integrator 和当前任务真正需要的执行/审查 Roles。
- 包含 Owner、Needs、Produces、Consumer、Writes、Acceptance 的 Task Graph。
READY / BLOCKED / HANDOFF / REVIEW协作事件和决策升级边界。forming → briefing → executing → reviewing → integrating → complete/stopped生命周期。
实际进入 Team 模式时读取 references/team-protocol.md，按其中的 Charter、角色、任务和阶段合同执行。需要一个完整示例时读取 references/development-team-example.md。
3. 映射到原生运行时
- Codex：Team 默认是
hub-and-spoke；当前主线程是 Lead，原生 subagents/agent threads 是 Members。不要假设 Members 可以直接互相通信。 - Claude Code：中心化 Team 使用 subagents；只有 Members 必须互相消息、认领共享任务或协作决策时才使用 Agent Teams。
- 不从 Codex shell 启动 Claude 充当 Member，也不从 Claude shell 启动 Codex。
准备创建 Member 时读取 references/native-runtimes.md，以当前宿主实际暴露的工具和能力为准，不凭记忆编造命令、参数或会话能力。
4. 派发与协作
Lead 负责 Charter、Task Graph、成员选择、依赖、写入所有权、跨 Task 决策、Review、Integration 和最终汇报。
每个 Member 同时只有一个 active Task。派发必须对空白上下文自包含，至少说明：
- Role 和 Goal
- 必读的 3-6 个路径、符号或直接 Handoff
- Writes 和明确不负责的范围
- Needs、Produces 和 Consumer
- Acceptance 和完成 Handoff
不要复制完整用户聊天、全部计划、其他 Member 日志或无关仓库历史。若 spawn 接口暴露 fork_turns、fork_context 或类似选项，使用空白或最小上下文，并在派发消息中提供稳定任务事实。
依赖未满足的 Task 不进入 active。正常运行不频繁 ping；只有完成、阻塞、异常或需要决策时介入。结果偏离时优先在原 Member 上精确修正一次；同一控制失败两次时停止扩张团队，选择重新分配、降级 Solo 或报告阻塞。
5. 写入、Git 与权限
- 并行只读 Task 可以共享 checkout。
- 并行写 Task 必须使用独立 worktree，或拥有明确且不重叠的路径。
- 一个文件、生成物或迁移状态同一时刻只有一个 Owner。
- 记录已有 dirty worktree；任何 Member 不得覆盖、清理或回退用户及其他成员的改动。
- Members 默认不 commit、merge、rebase、打 Tag、推送或发布；只有 Task 合同明确授权时才允许。
- Integrator（默认 Lead）负责最终 diff、冲突处理、团队级验证和经用户授权的 Git/发布动作。
- 不启用
--yolo、bypass 或无 sandbox 模式。多 Agent 不扩大当前会话的授权边界。
What ships with it
13 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
- agents/openai.yaml 377 B
- init.sh 2.6 KB runs code
- references/development-team-example.md 5.2 KB
- references/native-runtimes.md 5.6 KB
- references/team-protocol.md 7.0 KB
- scripts/check-update.sh 3.1 KB runs code
- templates/DECISIONS.md 317 B
- templates/PLAN.md 574 B
- templates/PROJECT.md 719 B
- templates/STATUS.md 771 B
- templates/task.md 877 B
- templates/TEAM.md 1.6 KB
- VERSION 14 B
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 1mo ago Changed · -111 lines · -29 tokens per session de504f5b3e30
- 1mo ago First seen · 208 lines · 126 tokens per session scan A 12307db59cf8
agent-taskgraph is a skill published in the GitHub repository wu736139669/agent-taskgraph-protocol (4 stars, last pushed 9d ago), licensed Apache-2.0. It adds 97 tokens to every session and 1,739 once invoked, about $0.0004 per session on Opus 5.5. A static security scan graded it A with 0 findings. It is 100% identical to agent-taskgraph, differing in 0 lines, and is treated as a copy.
Other skills, from other repositories
fabric-swarm
Creates a self-organizing team of persistent Pi Fabric actors with durable topics, mailboxes, and compare-and-swap tasks. Use for messenger-like collaboration and long-lived delegated work.
dispatch
Use when a task file exists in .hyperflow/tasks/ and workers need dispatching. Fans out parallel workers under per-batch Reviewers, runs a final integration review, and commits per sub-task. Endpoint of the auto-chain — no auto-deploy. Trigger with /hyperflow:dispatch, "run the plan", "execute the task", "build it"…
fabric-jev
Supervise shell commands and bounded task output from Python host calls to Jev TypeScript programs, with explicit optional typed decisions. Use for foreground/background shell workflows, semantic routing, verification, and per-turn advisors without a reasoning-model turn per event.
fabric-foreman
Starts a bounded Jev foreman above Main's coding work, assessing progress, drift, blockers, and verification per turn or at agent settlement with deterministic intervention policy.
fabric-exec
Python-only troubleshooting and advanced host API reference for fabricexec. Routine pi. coding calls are documented by ambient guidance; load this skill only after an argument-shape error or when an advanced surface needs exact contracts.
fabric-fusion
Multi-model deliberation. Two to 8 distinct models answer in parallel with web-capable tools, then a judge compares consensus, contradictions, coverage gaps, unique insights, and blind spots. Act mode runs 1–4 read-only references, then one actor reconciles and executes. Use when the cost of being wrong justifies…