实测可用的 DeepSeek Harness 技能库：每个 SKILL.md 都通过格式校验与加载冒烟，复制即用（中文优先）
Pending human review — the facts below come from the source repository.
实测可用的 DeepSeek Harness 技能库：每个 SKILL.md 都通过格式校验与加载冒烟，复制即用（中文优先）
Users who want this capability in DSH; check the source and docs before installing.
dsh plugin --profile web add github:hackerFish/awesome-dsh-skills No supported platforms were declared.
@deepseek-ai/dsh-skill-filesystem rules plus an isolated-DSH_HOME load smoke test. Copy, drop in, done. 中文 · Sibling repos: dsh-lab · awesome-dsh-presets · dsh-video-studio · dsh-restart ## What DSH skills are DSH scans skill roots (~/.dsh/skills, .agents/skills, project .dsh/skills) and registers SKILL.md files into the model-visible catalog. Each skill: name (kebab-case) + description required in frontmatter; optional whenToUse / metadata / disable-model-invocation / user-invocable. ## Install bash git clone https://github.com/hackerFish/awesome-dsh-skills ~/dsh-skills mkdir -p ~/.dsh/skills cp -r ~/dsh-skills/skills/* ~/.dsh/skills/ # all, or copy individual skill dirs ## Quality gates (every skill) 1. node tools/validate-skills.mjs — field whitelist, kebab-case names, boolean value forms, single-level structure, per official rules 2. Load smoke test in an isolated DSH_HOME (skills placed, harness booted clean) 3. Content rule: verified facts only; no unverified "magic prompts" ## Skills (15) | Skill | Purpose | |---|---| | dsh-git-commit | Conventional commits + pre-commit self-check | | dsh-code-review | Six-dimension structured review | | dsh-test-first | Red→green→refactor discipline | | dsh-doc-sync | Keep docs/changelog/versions in sync | | dsh-plugin-dev | DSH plugin dev with verified manifest/patch structures | | dsh-dependency-audit | Dependency & install-script risk audit | | dsh-refactor-safe | Baseline-first safe refactoring | | dsh-debug-session | DSH boot/plugin failure triage | | dsh-changelog | Keep a Changelog discipline | | dsh-chinese-docs | Chinese technical writing conventions | | dsh-pr-review | PR review checklist & structured feedback | | dsh-task-breakdown | Task decomposition with acceptance criteria | | dsh-skill-adapter | Adapt public portable skills to DSH discovery roots, active tools, and safety constraints | | dsh-webapp-testing | Browser-evidence web-app testing with console, network, server, and regression-test checks | | dsh-office-artifacts | Format-aware XLSX, DOCX, PPTX, and PDF creation or repair with reopen/render verification | See CONTRIBUTING.md to submit a skill. Contributors: CONTRIBUTORS.md ## License MITNot by stars — picked by hand, one at a time.
A visual settings dashboard for DeepSeek Harness: every registered settings namespace — including third-party plugin ones the official UI never covers — rendered as editable, schema-driven forms. Early, but the right direction: stop hand-editing YAML.
A DIY trajectory visualizer for DeepSeek Harness: watch the agent’s plan and execution unfold as a graph instead of scrolling raw logs. Early, but it fills a real gap — understanding at a glance what the agent is doing.
Dispatch DSH work from Claude Code / Codex: spin up in-host DSH agent sessions with per-tier presets, watch native subagent progress, and borrow a multimodal bridge that lends the text-only harness vision and image generation. An orchestration plugin that turns DSH into a backend for other coding agents.