Portable Spec-programming Skill for coding agents: consistent requirements, decisions, code, tests and docs without a forced directory layout, on Codex / Claude Code / GitHub Copilot.
DSH integration
Ecosystem-related
Author-claimed
Safety audit
Unaudited
Last verified
2026-09-04
License
MIT
01What can it help you accomplish?
Keep requirements, decisions, implementation, tests and docs consistent while building features, fixing bugs, changing architecture, or removing code
Updated decision records, code, tests, generated files and current docs kept consistent in the same pull request or linked changes
Developers using Codex, Claude Code or GitHub Copilot who want Spec-driven discipline without adopting a fixed directory layout
Write a proposal or review a change without implementing it
A proposal comparing real alternatives and defining direct checks, or a read-only consistency report — without modifying files
Teams that need design proposals or independent review before touching code
02How to install into DeepSeek Harness
Prerequisites
- Node.js (required by the community `skills` installer, per README)
Installation steps
- 01
Run `npx skills add songyang0603/ds-spec-loop` to install for your chosen coding agent and scope (requires Node.js).
$ npx skills add songyang0603/ds-spec-loop
- 02
Run `gh skill install songyang0603/ds-spec-loop ds-spec-loop --agent codex --scope user`, replacing `codex` with `claude-code` or `github-copilot`.
- 03
Clone or download the repo and copy the entire `skills/ds-spec-loop` directory into the agent’s skills folder (user or repository scope).
Verify the integration
Not specified by the author
03DSH integration and capability boundaries
Ecosystem Skill: an independent community project informed by public DeepSeek Harness engineering patterns that reuses each repo’s existing Spec/decision/test/CI conventions without requiring the DeepSeek Harness runtime, directories, or framework.
Spec-consistent change loop
a feature, bug fix, architecture change, or code-removal request→updated decision records, code, tests, generated files, and current docs kept consistent in one change
reads and may modify repository files (instructions, proposals, decisions, source, tests, generated files, docs, CI) when keeping the repo consistentDecision record management
any non-mechanical change→one decision record created or updated to own that change
Convention-aware reuse (no forced layout)
a repository with its own Spec / decision / test / CI conventions→reuses existing files instead of imposing a new directory layout
04Who is it for? When not to use it?
Good for
- Developers using Codex, Claude Code or GitHub Copilot who want Spec-driven discipline without adopting a fixed directory layout
- Teams that need design proposals or independent review before touching code
Not for
- When keeping the repo consistent, the Skill can read and modify repository files (code, tests, docs, CI, decision records). Review-only and proposal-only modes stay read-only by default and must not silently start implementing.
05Compatibility, maintenance and safety notes
- When keeping the repo consistent, the Skill can read and modify repository files (code, tests, docs, CI, decision records). Review-only and proposal-only modes stay read-only by default and must not silently start implementing.
- Independent community project: not official, not affiliated with or endorsed by DeepSeek, and does not redistribute DeepSeek Harness source code (see NOTICE).
Independent community Skill; MIT; last update-log v0.2 (2026-08-17); not affiliated with DeepSeek.
06Frequently asked questions
What coding agents does ds-spec-loop support?
Codex, Claude Code and GitHub Copilot. Install with `npx skills add songyang0603/ds-spec-loop` or `gh skill install songyang0603/ds-spec-loop ds-spec-loop --agent codex --scope user`, or copy `skills/ds-spec-loop` manually into the agent’s skills folder.
Is ds-spec-loop official DeepSeek Harness software?
No. It is an independent community project informed by public DeepSeek Harness patterns. It does not require the DeepSeek Harness runtime, is not affiliated with or endorsed by DeepSeek, and does not redistribute DeepSeek Harness source code.
Does it force a specific repository layout?
No. The Skill finds which existing files already do each job (AGENTS.md, ADRs, tests, CI) and reuses them. A Python project is not told to run `pnpm test`, and a repo that uses ADRs is not told to create another decision directory.
Can it review or propose without changing code?
Yes. Review mode checks whether proposals, decisions, code, tests, generated files and docs agree and stays read-only; proposal mode writes a proposal comparing real alternatives without implementing it.
07Related DSH workflows
distilly
by titanwings
Meta-skill turning source material into a versioned Person Profile skill with incremental merge, correction layer, rollback, per-host install.
dsh-market
by dsh-market
插件市场 (plugin marketplace): 打开 Settings → Plugin Market 浏览/搜索/一键安装社区插件；安装源限制为 curated awesome-dsh-plugin 注册表（其余拒绝），默认禁用构建脚本（pnpm>=10 需显式开启）。
ai_animation
by unclecheng-li
Generate standalone animated HTML diagrams, presentations, notes, protocol visualizations, and UI demos from prompts.
aegis
by ganyuanran
Make AI coding agents architecture-aware: baseline-first, evidence-verified, drift-checked, and safe across long tasks.
08Data and sources
This independent community project was developed from analysis of Spec and decision practices visible in the public [Dee…
This page is generated from the project’s public documentation, repository metadata and a structured parse of DSH Plugins; last verified on 2026-09-04. Found an error? Submit a correction.
Best DeepSeek Harness Plugins
Twelve plugins worth installing first — picked from the whole catalog, across every category.