Dsh Subagent Admission — Development skill for Claude Code
Shared lifecycle admission protocol and reference policy kernel for DeepSeek Harness subagents.
How to install Dsh Subagent Admission
This entry records only its repository, not the path inside it, so there is no
exact command to give. Open yha9806/dsh-subagent-admission and copy the folder into
~/.claude/skills/, or the file into ~/.claude/agents/.
What Dsh Subagent Admission does
Shared lifecycle admission protocol and reference policy kernel for DeepSeek Harness subagents.
Alternatives in Development
- IPolloWork — Enterprise-grade, local-first Agent Workbench for people and agent teams 4.9k ★
- Petdex — A public gallery of animated pets for Codex, Claude Code, DeepSeek Harness, Hermes, OpenCode, Gemini CLI, and 4k ★
- Flowix — Notes for you, Memory for your agents 374 ★
README
dsh-subagent-admission
[简体中文](README.zh-CN.md)
[](https://github.com/yha9806/dsh-subagent-admission/actions/workflows/ci.yml)
An experimental shared lifecycle admission protocol and reference policy kernel for [DeepSeek Harness](https://github.com/deepseek-ai/deepseek-harness) subagents.
This project addresses the failure shape reported in [Discussion #131](https://github.com/deepseek-ai/deepseek-harness/discussions/131): nested subagents can expand until one Web host becomes unresponsive. It does not add another orchestrator. It makes the decision to create or reactivate a subagent explicit, atomic, and lifecycle-owned across every caller that uses the official runtime.
The long-term product direction is an **Agent Runtime Resource Control Plane**: every materialising agent workload should first acquire an explicit, observable, lifecycle-owned resource permit. v0.1 deliberately tests only the DSH subagent boundary.
**Release-candidate status:** GitHub prerelease [`v0.1.0-rc.1`](https://github.com/yha9806/dsh-subagent-admission/releases/tag/v0.1.0-rc.1) is public, while the package is not published to npm. The focused upstream extension-point question is now public in [Discussion #131](https://github.com/deepseek-ai/deepseek-harness/discussions/131#discussioncomment-18020293). This remains an independent community project, not an official DeepSeek component. No DeepSeek endorsement or adoption, external-user evidence, production deployment, or maintainer response has been observed.
Maintainer review path
The upstream decision is deliberately one question:
Would an optional lifecycle-owned subagent admission capability belong as a documented extension point?
A bounded review can follow four artifacts:
- Agent Note: the protocol-v1 Service Definition, Provider, Consumer, failure, and ownership contract;
- [slim
Related Skills
Maestro Preset
Multi-agent orchestration preset for the DeepSeek Harness (DSH): cross-plane supervisor over Orca / dais (ex-z
Dsh Acp Plugin
Agentic Control Plane for DeepSeek Harness — policy-check every tool call before it runs
Unified Agent Memory
Unified fleet-wide agent memory system for DeepSeek Harness — shared Obsidian Vault for dsh, Codex, Claude Cod
Dsh Agent Hub
DeepSeek Harness host plugin: Agent Hub mailbox tools shared by deepseek / claude-code / codex
Dsh Agent Mailbox
DeepSeek Harness host plugin: shared file mailbox tools for Agent Hub (deepseek / claude-code / codex)
Chinese Script Policy
Harness-neutral Traditional Chinese enforcer and offline converter (skill / DSH bundle / CLI). Checks four axe
Related Agents
Ds Flash
DSH (DeepSeek Harness) worker on the flash tier - fast and cheap. Delegate simple, well-scoped subtasks: mecha
Caio
Chief AI Officer — AI strategy, model lifecycle governance, EU AI Act / NIST AI RMF posture, evaluation harnes
Harness Eng
Use for every build task on a deepseek-harness core — one layer per claimed packet, gated by hedgehog verify.