DecimalAI Skills Registry

ai.decimal由 decimal.ai 发布v0.1.3更新于 Oct 24, 2026

Search agent skills ranked by measured lift vs a no-skill baseline, with safety and rankings.

已验证STDIO仅桌面知识与记忆开发者工具AI 与机器学习

概览

AI 生成的概览

面向 DecimalAI 公共技能注册表的只读搜索与排行榜工具，按实测基准效果对智能体技能排序。

可以做什么
通过公共 DecimalAI 技能注册表提供三个只读工具：search_skills 进行关键词与语义混合搜索；get_skill 返回完整记录，包括安全扫描状态、经验证的基准提升、SkillScore、评分以及 SKILL.md 正文；get_leaderboard 按 skill_score、biggest_improvement、efficiency 或 top_rated 返回排名列表。三个工具都读取公共端点，可匿名使用。设置 API 密钥只会增加按组织维度的补充信息，例如 installed_as。
适用场景
当助手需要按实测效果而非下载量来查找、比较或引用智能体技能时适用，例如为某项任务挑选技能，或在采用前查看技能的安全扫描状态与基准提升。
运行要求
以 PyPI 包 decimalai-mcp 通过 stdio 在本地运行，需要 Python 3.10+（pip install 或 uvx）。需要访问公共注册表端点的网络连接。无需账号或 API 密钥；可选密钥 DECIMAL_API_KEY 用于启用按组织维度的补充信息，DECIMAL_API_URL 可指向自托管后端。
安装前请注意
该服务器为只读且无需密钥，匿名使用时只会把搜索查询发送到公共注册表。若设置 DECIMAL_API_KEY，该凭据会传给服务器并使请求与你的组织关联；请将其视为机密，在不需要按组织补充信息时不要设置。

安装

在 SourceWeft 中

  1. 打开 控制台中的 DecimalAI Skills Registry，将其添加到工作区。
  2. 为需要使用其工具的对话启用该服务。

Desktop only，通过 STDIO。 STDIO 服务会启动本地进程，因此需要 SourceWeft 桌面宿主。

其他 MCP 客户端

参照 仓库 中的启动说明。

README

decimalai-mcp

Part of DecimalAI. Most users want the Python SDK → decimal-labs/decimalai-python.

MCP server for the DecimalAI skills registry — the registry that ranks agent skills by measured effectiveness (verified A/B benchmarks, live pass rates, AI rater scores), not download counts.

[PyPI] [Downloads] [CI] [Python] [License]

Gives any MCP client (Claude Desktop, Claude Code, Cursor, …) three read-only tools:

ToolWhat it does
search_skills(query, category?, sort?, limit?)Hybrid keyword/semantic search over the public registry
get_skill(slug)Full record: trust & safety-scan status, verified benchmark lift, SkillScore, ratings, SKILL.md body, and how to install it (decimalai skills add <slug>, the folder it lands in, and the Claude Code --plugin-dir line)
get_leaderboard(sort?, category?, window_days?, limit?)Ranked leaderboard: skill_score, biggest_improvement, efficiency, top_rated

No API key required — all three tools read public registry endpoints. If you set DECIMAL_API_KEY (from app.decimal.ai/settings), the same tools additionally show which skills your org has already installed (installed_as).

Install

bash
pip install decimalai-mcp# or, no install needed at config time:uvx decimalai-mcp

Requires Python 3.10+.

Claude Code

bash
claude mcp add decimalai -- uvx decimalai-mcp# with an API key:claude mcp add decimalai -e DECIMAL_API_KEY=dai_sk_... -- uvx decimalai-mcp

Claude Desktop

Add to claude_desktop_config.json (Settings → Developer → Edit Config):

json
{  "mcpServers": {    "decimalai": {      "command": "uvx",      "args": ["decimalai-mcp"],      "env": {        "DECIMAL_API_KEY": "dai_sk_optional"      }    }  }}

Omit the env block entirely for anonymous read-only access.

Configuration

Env varDefaultPurpose
DECIMAL_API_KEY(unset)Optional. Unlocks per-org enrichment (e.g. installed_as) on the same public endpoints.
DECIMAL_API_URLhttps://api.decimal.aiPoint at a self-hosted / local backend.

Why no check_manifest_impact tool?

The manifest-impact endpoint (POST /api/v1/regression-check) is authenticated on the platform — it analyzes your org's production traces against a candidate manifest, so there is no public variant to expose. This server is deliberately a read-only, key-optional public-registry surface. If demand shows up, an authed check_manifest_impact (requiring DECIMAL_API_KEY) is a natural v0.2 addition; the regression-check GitHub Action covers the CI use-case today.

Endpoints used (all public)

  • GET /api/v1/registry/skills — browse/search
  • GET /api/v1/registry/skills/{slug} — detail
  • GET /api/v1/registry/leaderboard — ranked leaderboard (category filtering falls back to the browse endpoint's documented view=ranks mode, because the leaderboard endpoint is uncategorized)

Development

bash
pip install -e ".[dev]"pytest              # all HTTP mocked; no networkpython -m decimalai_mcp.server   # run over stdio

Run pytest yourself before opening a PR. CI also asserts that the pinned mcp<2 still provides FastMCP, which the mocked tests do not cover — run that one too:

bash
python -c "from mcp.server.fastmcp import FastMCP; import decimalai_mcp.server"

Releases are cut from a published GitHub Release — see RELEASING.md for the gates a change has to pass and the version strings that must move together.

License

MIT

来源：README.md，提交 a033c9f

工具

0
SourceWeft 尚未在沙箱中启动过该软件包，因此其工具未知。

版本历史

1
  1. v0.1.3最新Sep 16, 2026

评分

已满足适用检查项的 50%

该评分衡量此收录条目的完整程度，依据注册中心条目、端点与公开来源计算得出。它不是对服务器本身的评价，也不表示服务器安全或当前可用：可连通性是带日期的检查结果，评分也不包含安装量或用户评价。

  • 近 30 天内可连通25 分

    请保持端点能够响应 MCP 握手；SourceWeft 每 30 天探测一次。

  • 已声明认证方式10 分

    请在注册中心条目中声明凭据请求头，或保持端点能够响应探测，以便判定其认证方式。

  • 已列出工具15 分

    请让服务器在无凭据时响应 tools/list，或发布可在沙箱中无凭据启动的软件包。

  • 描述必需15 分
  • 图标10 分
  • 文档15 分
  • 许可证10 分