Use it when adding agent definitions, managing foreground or background tasks, restoring interrupted runs, verifying results, or handling isolated worktrees.

It keeps delegated work recoverable and isolated, including when a process stops or the parent agent is missing a result.

knowledge-agent-execution-and-orchestration-subagen

Q: What is knowledge-agent-execution-and-orchestration-subagent-runtime?

A runtime for defining, running, pausing, resuming, and verifying child coding agents working on separate tasks.

Install

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.

Any agent

npx skills add echoVic/blade-code --skill subagent-runtime

Clone the repo

git clone --depth 1 https://github.com/echoVic/blade-code

Made for: Claude Code, Codex.

Wrote 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.

README.md

[![agentmods](https://agentmods.dev/badge/skills/echovic/blade-code/subagent-runtime/github.svg)](https://agentmods.dev/skills/echovic/blade-code/subagent-runtime)

Your own site

<a href="https://agentmods.dev/skills/echovic/blade-code/subagent-runtime"><img src="https://agentmods.dev/badge/skills/echovic/blade-code/subagent-runtime/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.

Your own site · 80×15

<a href="https://agentmods.dev/skills/echovic/blade-code/subagent-runtime"><img src="https://agentmods.dev/badge/skills/echovic/blade-code/subagent-runtime.svg" alt="Reviewed on agentmods" width="80" height="20"></a>

Per session 117 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.

When invoked 2,367 The whole file, excluding the scripts and references it only reads on demand.

Security scan A 0 findings. A grade says what 26 rules found in the file — not that it is safe. Third-party audits

NVIDIA SkillSpector pass 7 Sept 2026

How audits are shown

Origin original No closer match found in the catalogue.

Token cost

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	$0.00117	$0.02367
Opus 5.5	$0.00047	$0.00947
Sonnet 5.5	$0.00023	$0.00473
Haiku 4.5	$0.00012	$0.00237

Measured 1mo ago against content hash 40f55a36e2d1, method: parsed. Prices are Anthropic first-party input rates as of 2026-10-07, from the pricing page.

Security

Grade A, and why

knowledge-agent-execution-and-orchestration-subagent-runtime 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.

.trae/knowledges/agent-execution-and-orchestration/subagent-runtime/SKILL.md · 71 lines

How it starts

The opening of the file, as written. The whole thing — 71 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Module Structure

子代理是拥有独立 SessionRuntime、transcript 和 sidecar 的 child run；Task 工具负责身份与调用契约，执行器负责运行，后台管理器负责持久状态、通知和崩溃恢复。

Directory Layout

packages/cli/src/agent/subagents/ — 注册表、执行器、sidecar、恢复、结果采纳和内置代理
- SubagentRegistry.ts — Markdown/frontmatter 配置加载和优先级
- SubagentExecutor.ts — 独立 child SessionRuntime 执行
- BackgroundAgentManager.ts — 后台运行、等待、取消、resume 和 orphan reconciliation
- AgentSessionStore.ts — durable sidecar 与 lineage/config snapshot
- SubagentWorktreeLifecycle.ts — 隔离 worktree 准备与收尾
- SubagentResultAdoption.ts — 父 Task 缺失结果时的重启采纳
- BackgroundSubagentCompletion.ts — 后台终态通知构造
packages/cli/src/tools/builtin/task/task.ts — Task 工具前台、后台和 resume 协议
packages/cli/src/cli/agents.ts — --agents 单次运行定义严格解析
packages/cli/src/slash-commands/agents.ts — TUI agent 管理入口
.claude/agents/ — Git 可见的 Claude Code 兼容 agent 定义
docs/guides/subagents.md — 用户可见配置与恢复契约

Key Entry Points

createTaskTool() in packages/cli/src/tools/builtin/task/task.ts — 模型侧委派入口
SubagentRegistry.loadFromStandardLocations() in packages/cli/src/agent/subagents/SubagentRegistry.ts — 分层资源发现
SubagentExecutor.execute() in packages/cli/src/agent/subagents/SubagentExecutor.ts — 直接前台执行器
BackgroundAgentManager.startBackgroundAgent() in packages/cli/src/agent/subagents/BackgroundAgentManager.ts — durable 后台 child 启动
BackgroundAgentManager.resumeAgent() in packages/cli/src/agent/subagents/BackgroundAgentManager.ts — immutable child resume

Gotchas

resume_from 不会复用原 agent ID，而会创建新的 child run；源运行必须已终止、类型必须一致，并记录 rootAgentId → resumedFrom → resumeDepth，否则恢复被拒绝 (packages/cli/src/tools/builtin/task/task.ts, packages/cli/src/agent/subagents/BackgroundAgentManager.ts)
resume 使用源 sidecar 中冻结的模型、权限、工具、系统提示、最大回合和隔离配置，不重新读取同名 agent 的最新定义；修改配置不会改变既有 lineage (packages/cli/src/agent/subagents/AgentSessionStore.ts, packages/cli/src/agent/subagents/BackgroundAgentManager.ts, git:888d608f)
子代理访问权按 parent sessionId + canonical projectPath 复合校验；只有相同 session ID 不足以跨 workspace 查询、取消、采纳或恢复 child (packages/cli/src/agent/subagents/AgentSessionStore.ts, packages/cli/src/tools/builtin/task/task.ts)
子 Agent 的 blacklist 总会加入 EnterWorktree、ExitWorktree 和 TeamCreate；隔离 worktree 由宿主生命周期创建，teammate 也不能嵌套创建团队 (packages/cli/src/agent/subagents/SubagentExecutor.ts, packages/cli/src/agent/subagents/BackgroundAgentManager.ts)
隔离运行只有“成功且无改动、无提交”才自动删除 worktree；失败或任何变更都会保留路径与分支，调用方必须按交付语义处理 (packages/cli/src/agent/subagents/SubagentWorktreeLifecycle.ts)
后台完成通知只内联最多 32,000 字符结果和 8,000 字符错误，并明确标为不可信数据；需要完整 durable 结果时才调用 TaskOutput，不能通过高频轮询代替 completion push (packages/cli/src/agent/subagents/BackgroundSubagentCompletion.ts, packages/cli/src/tools/builtin/task/task.ts)
orphan child 只有在 durable turn 已 turn_completed 且恢复上下文中存在最终 assistant 文本时才恢复为 completed；其他情况记为 interrupted failed 并允许用新 ID resume (packages/cli/src/agent/subagents/BackgroundAgentManager.ts)
running sidecar 同时保存 PID 和进程启动身份，防止 PID 复用被误判为原 owner 仍存活；旧记录缺身份时采用更保守的兼容窗口 (packages/cli/src/agent/subagents/AgentSessionStore.ts, packages/cli/src/agent/subagents/BackgroundAgentManager.ts)
父 transcript 中缺失前台 Task 结果时，只有 child ID、owner、description、显式 type、resume lineage 和终态结果全部匹配才会采纳；后台 child 或 cancelled child 不走该路径 (packages/cli/src/agent/subagents/SubagentResultAdoption.ts, git:103e57b7)
verification 与 goal-verification 是保留名称，用户、项目和命令行 override 都不能覆盖；普通内置 agent 则允许被后续高优先级来源替换 (packages/cli/src/agent/subagents/SubagentRegistry.ts)

Read the full file on GitHub · 71 lines

Changes

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 First seen · 71 lines · 117 tokens per session scan A 40f55a36e2d1

Subscribe to this mod's changes

knowledge-agent-execution-and-orchestration-subagent-runtime is a skill published in the GitHub repository echoVic/blade-code (181 stars, last pushed 3d ago), licensed MIT. It adds 117 tokens to every session and 2,367 once invoked, about $0.0005 per session on Opus 5.5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-09-03.

Building agents Git and workflow LLM agent ai cli code mcp

pypi-release

This skill should be used when releasing tunacode-cli to PyPI. It keeps the existing local release checks, then hands the actual PyPI upload to a GitHub Actions workflow that uses the repository's PYPIAPITOKEN secret.

alchemiststudiosDOTai/tunacode · 52 tokens

datalad

Retrieve, version, and publish scientific datasets with DataLad and git-annex, and capture computational provenance with datalad run, rerun, and containers-run. Use when cloning or fetching data from OpenNeuro, DANDI, datasets.datalad.org, or any DataLad dataset; when a file in a dataset reads as a broken symlink or a…

synthetic-sciences/openscience · 146 tokens

knowledge-agent-execution-and-orchestration-subagent-runtime