Pipelex：AI 工作流声明式编程语言

构建和运行 AI Method

Method 是一个可复用的、类型化的 AI 过程 —— 在 .mthds 文件中声明，由 Pipelex 执行。每个步骤都是显式的，每个输出都是结构化的，每次运行都是可重复的。

Method 的外观

[pipe.summarize_article]
type    = "PipeLLM"
inputs  = { article = "Text", audience = "Text" }
output  = "Text"
prompt  = "Summarize $article in three bullet points for $audience."

从这里开始，Pipelex 处理 60+ 个模型的路由、结构化输出解析和管道编排。

使用 Claude Code（推荐）

将以下内容粘贴到 Claude Code：

Install mthds: run bash -c 'npm install -g mthds && mthds-agent bootstrap && claude plugin marketplace add mthds-ai/mthds-plugins && claude plugin install mthds@mthds-plugins'. Then tell me to /reload-plugins.

构建你的第一个 method：

/mthds-build A method to summarize articles with key takeaways for different audiences

/mthds-run

在 Codex 中使用

将以下内容粘贴到 Codex：

Install mthds: run bash -c 'npm install -g mthds && mthds-agent bootstrap && mthds-agent codex install-hook && mthds-agent codex apply-config && codex plugin marketplace add mthds-ai/mthds-plugins'. Then tell me to restart Codex and run /plugins to install mthds.

为 .mthds 语法高亮和流程图可视化安装 VS Code 扩展。

npm install -g mthds
mthds-agent bootstrap
pipelex init

为 .mthds 语法高亮和流程图可视化安装 VS Code 扩展。

验证一切设置正确：

pipelex doctor

如果你只需要 Pipelex CLI 而不需要 agent 集成：

uv tool install pipelex
pipelex init

Pipelex Gateway（推荐）

Pipelex Gateway 提供免费额度、统一的 LLM API 密钥、跨所有主要提供商的 OCR / 文档提取和图像生成。获取你的密钥，将 PIPELEX_GATEWAY_API_KEY=your-key-here 添加到 ~/.pipelex/.env，运行 pipelex init。

自带密钥

使用来自 OpenAI、Anthropic、Google、Mistral 等的现有 API 密钥。详见配置 AI 提供商。

本地 AI

Ollama、vLLM、LM Studio 或 llama.cpp —— 无需 API 密钥。详见配置 AI 提供商。

生产级示例：批量 CV 筛选

一个生产级 method，接受一批 CV 和招聘 PDF，提取并分析每一份，然后评分每个候选人与职位的匹配度。

cv_batch_screening.mthds

[pipe.batch_analyze_cvs_for_job_offer]
type = "PipeSequence"
description = """
Main orchestrator pipe that takes a bunch of CVs and a job offer in PDF format, and analyzes how they match.
"""
inputs = { cvs = "Document[]", job_offer_pdf = "Document" }
output = "CandidateMatch[]"
steps = [
  { pipe = "prepare_job_offer", result = "job_requirements" },
  { pipe = "process_cv", batch_over = "cvs", batch_as = "cv_pdf", result = "match_analyses" },
]

[concept.CandidateProfile]
description = "A structured summary of a job candidate's professional background extracted from their CV."

[concept.CandidateProfile.structure]
skills       = { type = "text", description = "Technical and soft skills possessed by the candidate", required = true }
experience   = { type = "text", description = "Work history and professional experience", required = true }
education    = { type = "text", description = "Educational background and qualifications", required = true }
achievements = { type = "text", description = "Notable accomplishments and certifications" }

[concept.JobRequirements]
description = "A structured summary of what a job position requires from candidates."

[concept.JobRequirements.structure]
required_skills  = { type = "text", description = "Skills that are mandatory for the position", required = true }
responsibilities = { type = "text", description = "Main duties and tasks of the role", required = true }
qualifications   = { type = "text", description = "Required education, certifications, or experience levels", required = true }
nice_to_haves    = { type = "text", description = "Preferred but not mandatory qualifications" }

[concept.CandidateMatch]
description = "An evaluation of how well a candidate fits a job position."

[concept.CandidateMatch.structure]
match_score        = { type = "number", description = "Numerical score representing overall fit percentage between 0 and 100", required = true }
strengths          = { type = "text", description = "Areas where the candidate meets or exceeds requirements", required = true }
gaps               = { type = "text", description = "Areas where the candidate falls short of requirements", required = true }
overall_assessment = { type = "text", description = "Summary evaluation of the candidate's suitability", required = true }

[pipe.prepare_job_offer]
type = "PipeSequence"
description = """
Extracts and analyzes the job offer PDF to produce structured job requirements.
"""
inputs = { job_offer_pdf = "Document" }
output = "JobRequirements"
steps = [
  { pipe = "extract_one_job_offer", result = "job_offer_pages" },
  { pipe = "analyze_job_requirements", result = "job_requirements" },
]

[pipe.extract_one_job_offer]
type        = "PipeExtract"
description = "Extracts text content from the job offer PDF document"
inputs      = { job_offer_pdf = "Document" }
output      = "Page[]"
model       = "@default-text-from-pdf"

[pipe.analyze_job_requirements]
type = "PipeLLM"
description = """
Parses and summarizes the job requirements from the extracted job offer content, identifying required skills, responsibilities, qualifications, and nice-to-haves
"""
inputs = { job_offer_pages = "Page" }
output = "JobRequirements"
model = "$writing-factual"
system_prompt = """
You are an expert HR analyst specializing in parsing job descriptions. Your task is to extract and summarize job requirements into a structured format.
"""
prompt = """
Analyze the following job offer content and extract the key requirements for the position.

@job_offer_pages
"""

[pipe.process_cv]
type = "PipeSequence"
description = "Processes one application"
inputs = { cv_pdf = "Document", job_requirements = "JobRequirements" }
output = "CandidateMatch"
steps = [
  { pipe = "extract_one_cv", result = "cv_pages" },
  { pipe = "analyze_one_cv", result = "candidate_profile" },
  { pipe = "analyze_match", result = "match_analysis" },
]

[pipe.extract_one_cv]
type        = "PipeExtract"
description = "Extracts text content from the CV PDF document"
inputs      = { cv_pdf = "Document" }
output      = "Page[]"
model       = "@default-text-from-pdf"

[pipe.analyze_one_cv]
type = "PipeLLM"
description = """
Parses and summarizes the candidate's professional profile from the extracted CV content, identifying skills, experience, education, and achievements
"""
inputs = { cv_pages = "Page" }
output = "CandidateProfile"
model = "$writing-factual"
system_prompt = """
You are an expert HR analyst specializing in parsing and summarizing candidate CVs. Your task is to extract and structure the candidate's professional profile into a structured format.
"""
prompt = """
Analyze the following CV content and extract the candidate's professional profile.

@cv_pages
"""

[pipe.analyze_match]
type = "PipeLLM"
description = """
Evaluates how well the candidate matches the job requirements, calculating a match score and identifying strengths and gaps
"""
inputs = { candidate_profile = "CandidateProfile", job_requirements = "JobRequirements" }
output = "CandidateMatch"
model = "$writing-factual"
system_prompt = """
You are an expert HR analyst specializing in candidate-job fit evaluation. Your task is to produce a structured match analysis comparing a candidate's profile against job requirements.
"""
prompt = """
Analyze how well the candidate matches the job requirements. Evaluate their fit by comparing their skills, experience, and qualifications against what the position demands.

@candidate_profile

@job_requirements

Provide a comprehensive match analysis including a numerical score, identified strengths, gaps, and an overall assessment.
"""

flowchart LR
    %% Pipe and stuff nodes within controller subgraphs
    subgraph sg_n_8b2136e3fe["batch_analyze_cvs_for_job_offer"]
        subgraph sg_n_91d5d6dc7c["prepare_job_offer"]
            n_fde22777cb["analyze_job_requirements"]
            s_f9f703fbb4(["job_requirements<br/>JobRequirements"]):::stuff
            n_b8469c838f["extract_one_job_offer"]
            s_d998350046(["job_offer_pages<br/>Page"]):::stuff
        end
        subgraph sg_n_f8d5afb7cd["process_cv_batch"]
            subgraph sg_n_6e53e16369["process_cv"]
                n_c18aded200["analyze_match"]
                s_5c911f7e54(["match_analysis<br/>CandidateMatch"]):::stuff
                n_a7ed00ac24["analyze_one_cv"]
                s_c5ae714e89(["candidate_profile<br/>CandidateProfile"]):::stuff
                n_d24f39aa60["extract_one_cv"]
                s_427beb5195(["cv_pdf<br/>Document"]):::stuff
                s_f1f80289df(["cv_pages<br/>Page"]):::stuff
            end
            subgraph sg_n_2cfb7a32c8["process_cv"]
                n_f6a25d1769["analyze_match"]
                s_ea99eee6ed(["match_analysis<br/>CandidateMatch"]):::stuff
                n_f48b73fbee["analyze_one_cv"]
                s_e1ffee913e(["candidate_profile<br/>CandidateProfile"]):::stuff
                n_d16f2fe381["extract_one_cv"]
                s_041bb18fb4(["cv_pdf<br/>Document"]):::stuff
                s_5fbba7194a(["cv_pages<br/>Page"]):::stuff
            end
            subgraph sg_n_08a7186be9["process_cv"]
                n_937e750ea4["analyze_match"]
                s_bb41a103f0(["match_analysis<br/>CandidateMatch"]):::stuff
                n_786a2969d5["analyze_one_cv"]
                s_c47fe821d7(["candidate_profile<br/>CandidateProfile"]):::stuff
                n_38f0cfd11c["extract_one_cv"]
                s_2634ece93d(["cv_pdf<br/>Document"]):::stuff
                s_44e253b325(["cv_pages<br/>Page"]):::stuff
            end
        end
    end

    %% Pipeline input stuff nodes (no producer)
    s_9b7e74ac51(["job_offer_pdf<br/>Document"]):::stuff

    %% Data flow edges: producer -> stuff -> consumer
    n_a7ed00ac24 --> s_c5ae714e89
    n_b8469c838f --> s_d998350046
    n_f48b73fbee --> s_e1ffee913e
    n_d16f2fe381 --> s_5fbba7194a
    n_fde22777cb --> s_f9f703fbb4
    n_d24f39aa60 --> s_f1f80289df
    n_38f0cfd11c --> s_44e253b325
    n_786a2969d5 --> s_c47fe821d7
    n_c18aded200 --> s_5c911f7e54
    n_f6a25d1769 --> s_ea99eee6ed
    n_937e750ea4 --> s_bb41a103f0
    s_c5ae714e89 --> n_c18aded200
    s_9b7e74ac51 --> n_b8469c838f
    s_d998350046 --> n_fde22777cb
    s_e1ffee913e --> n_f6a25d1769
    s_427beb5195 --> n_d24f39aa60
    s_041bb18fb4 --> n_d16f2fe381
    s_2634ece93d --> n_38f0cfd11c
    s_5fbba7194a --> n_f48b73fbee
    s_f9f703fbb4 --> n_c18aded200
    s_f9f703fbb4 --> n_f6a25d1769
    s_f9f703fbb4 --> n_937e750ea4
    s_f1f80289df --> n_a7ed00ac24
    s_44e253b325 --> n_786a2969d5
    s_c47fe821d7 --> n_937e750ea4

    %% Batch edges: list-item relationships
    s_52d84618d0(["match_analyses<br/>CandidateMatch"]):::stuff
    s_5c911f7e54 -."[0]".-> s_52d84618d0
    s_ea99eee6ed -."[1]".-> s_52d84618d0
    s_bb41a103f0 -."[2]".-> s_52d84618d0

    %% Style definitions
    classDef failed fill:#ffcccc,stroke:#cc0000
    classDef stuff fill:#fff3e6,stroke:#cc6600,stroke-width:2px
    classDef controller fill:#e6f3ff,stroke:#0066cc

    %% Subgraph depth-based coloring
    style sg_n_08a7186be9 fill:#fffde6
    style sg_n_2cfb7a32c8 fill:#fffde6
    style sg_n_6e53e16369 fill:#fffde6
    style sg_n_8b2136e3fe fill:#e6f3ff
    style sg_n_91d5d6dc7c fill:#ffe6e6
    style sg_n_f8d5afb7cd fill:#e6ffe6

运行示例

pipelex run bundle cv_batch_screening.mthds --inputs inputs.json

创建 inputs.json 文件，包含你的 PDF URL：

{
  "cvs": {
    "concept": "native.Document",
    "content": [
      { "url": "https://pipelex-web.s3.amazonaws.com/demo/John-Doe-CV.pdf" },
      { "path": "inputs/Jane-Smith-CV.pdf" }
    ]
  },
  "job_offer_pdf": {
    "concept": "native.Document",
    "content": {
      "url": "https://pipelex-web.s3.amazonaws.com/demo/Job-Offer.pdf"
    }
  }
}

Python 中的运行

import asyncio
from pipelex.core.stuffs.document_content import DocumentContent
from pipelex.pipelex import Pipelex
from pipelex.pipeline.runner import PipelexRunner
# Generated by: `pipelex build structures bundle cv_batch_screening.mthds`
from structures.cv_batch_screening__candidate_match import CandidateMatch

async def run_pipeline() -> list[CandidateMatch]:
    runner = PipelexRunner()
    response = await runner.execute_pipeline(
        pipe_code="batch_analyze_cvs_for_job_offer",
        inputs={
            "cvs": {
                "concept": "Document",
                "content": [
                    DocumentContent(url="https://pipelex-web.s3.amazonaws.com/demo/John-Doe-CV.pdf"),
                    DocumentContent(path="inputs/Jane-Smith-CV.pdf"),
                ],
            },
            "job_offer_pdf": {
                "concept": "native.Document",
                "content": DocumentContent(url="https://pipelex-web.s3.amazonaws.com/demo/Job-Offer.pdf"),
            },
        },
    )
    pipe_output = response.pipe_output
    print(pipe_output)
    return pipe_output.main_stuff_as_items(item_type=CandidateMatch)

Pipelex.make()
asyncio.run(run_pipeline())

见证 Pipelex 的威力

Claude Code 构建你的 AI Method

我们强烈推荐在你的 IDE 中安装我们的 .mthds 语法高亮扩展：

VS Code：从 VS Code Marketplace 安装

Cursor、Windsurf 和其他 VS Code fork：从 Open VSX Registry 安装，或在你的扩展选项卡中直接搜索 "Pipelex"

运行 pipelex init 也会在检测到你的 IDE 时自动提供安装扩展的选项。

同一个 .mthds 文件可以从多个执行目标运行：

从 TypeScript 使用 Pipelex

对于 Node、Next.js 或任何 TypeScript 应用，通过 mthds npm SDK（源码：mthds-js）调用 Pipelex API 服务器。自托管开源 pipelex-api 并将 SDK 指向你的实例。Pipelex 托管的运行器 api.pipelex.com 也在私测中 —— 加入等待列表。

npm install mthds

快速开始的最快方式：fork pipelex-starter-js 模板 —— 一个包含三个工作示例的 Next.js 16 + TypeScript 应用（文本实体提取、PDF 摘要、图像生成）。在 GitHub 上点击 Use this template。

在我们的 Cookbook 仓库中探索真实示例：

克隆它、fork 它，并用各种用例的生产级 method 进行实验。

该包支持以下附加功能：

anthropic：Anthropic/Claude 文本生成支持

google：Google 模型（Vertex）文本生成支持

mistralai：Mistral AI 文本生成和 OCR 支持

bedrock：Amazon Bedrock 文本生成支持

fal：Black Forest Labs "FAL" 服务的图像生成

linkup：Linkup 网络搜索

docling：Docling OCR

uv pip install "pipelex[anthropic,google,google-genai,mistralai,bedrock,fal,linkup,docling]"

隐私与遥测

Pipelex Gateway 仅收集技术数据（模型名称、令牌计数、延迟）—— 绝不收集 prompt 或业务数据。如果你想避免 Gateway 遥测，禁用 pipelex_gateway 并改用你自己的提供商密钥或本地 AI。了解更多

贡献

我们欢迎贡献！详见贡献指南。

社区

GitHub Issues · Discussions · Documentation

本项目采用 MIT 许可证。运行时依赖通过 PyPI 分发，采用各自的许可证。

"Pipelex" 是 Evotis S.A.S. 的商标。

© 2025-2026 Evotis S.A.S.