abl-mcp-server

abl-mcp-server

MCP server for OpenEdge ABL — pluggable tool architecture with per-project YAML configuration. Provides AI assistants with 24 tools to parse, analyze, lint, document, and scaffold ABL projects.

Category
访问服务器

README

abl-mcp-server

MCP (Model Context Protocol) server for OpenEdge ABL — pluggable tool architecture with per-project YAML configuration. Provides AI assistants with 24 tools to parse, analyze, lint, document, and scaffold ABL projects.

Built on:

Quick Start

npx github:breakit/abl-mcp-server

Or add to opencode config:

"abl": {
  "type": "local",
  "command": ["npx", "-y", "github:breakit/abl-mcp-server"],
  "enabled": true
}

Pluggable Architecture

Each tool is a separate module in src/tools/. Tools are auto-discovered at startup and can be enabled/disabled via a per-project YAML config file.

Per-Project Config (./abl-mcp-server.yaml)

Place this in your ABL project root:

tools:
  enabled:
    - read-abl-file
    - query-abl-symbols
    - analyze-dependencies
    - gen-doc-comment
    - gen-abldoc
    - gen-ablunit-test
    - abl-lint
    # ... add any tools you need
  disabled:
    - check-project-config

All tools are enabled by default. Add names to disabled to turn them off, or set enabled to a specific subset.

Adding Custom Tools

Drop a .ts file into ~/.config/abl-mcp-server/tools/:

import type { ToolModule } from '@breakit/abl-mcp-server/types'

export default {
  name: 'my-custom-tool',
  description: 'Does something custom',
  inputSchema: { type: 'object', properties: { input: { type: 'string' } }, required: ['input'] },
  handler: async ({ input }) => {
    return { content: [{ type: 'text', text: `Got: ${input}` }] }
  },
} satisfies ToolModule

Add my-custom-tool to your abl-mcp-server.yaml enabled list.

Tools (24 total, 13 enabled by default)

All 24 tools are available but only a curated subset is enabled by default. Enable additional tools via abl-mcp-server.yaml (see Pluggable Architecture above).

Default-enabled

Analytical

Tool Description
read-abl-file Parse an ABL file — list functions, includes, preprocessor defines
query-abl-symbols List all function symbols in a file
read-df-file Parse a .df schema — tables, fields, indexes, sequences
resolve-includes Resolve {include} paths against the project PROPATH
list-project-files List all .p/.w/.cls/.i files in a project
analyze-dependencies Build a full dependency graph — includes, calls, cycles, orphans
df-diff Compare two .df schema files — structured diff
find-dead-code Find unused functions, includes, and preprocessor defines
find-annotations Find TODO, FIXME, HACK, XXX, NOTE and similar marker comments
abl-lint Lint ABL files for coding conventions (37 rules defined in config.yaml)

Lint Rules

All 37 rules are defined in config.yaml (shipped with the server). They are inspired by Prolint. To see all active rules: call abl-lint with listRules: true.

Customize rules via your project's abl-mcp-server.yaml:

lint:
  rules:
    # Override severity of an existing rule
    no-undo:
      pattern: '^DEFINE (?:VARIABLE|VAR) +\w+ (?:AS \w+ )?(?!.*NO-UNDO)'
      message: 'DEFINE VARIABLE should include NO-UNDO'
      severity: warning

    # Add a custom rule
    my-naming-convention:
      pattern: '^\s*PROCEDURE\s+[a-z\d]'
      message: 'Procedure names should start with uppercase'
      severity: warning

    # Disable a rule by not including it in enabled (see tools.disabled pattern)
Group Rules
No-undo / Lock no-undo, no-undo-param
Deprecations pause, global-define, recid, shared
Shell / Security shell-call, hardcoded-email
Find / Performance no-lock-type, find-no-error, for-each-no-where, exclusive-no-wait, no-index
Style / Convention end-type, block-label, lex-colon, method-name-case, class-name-case, function-name-case, nolonglines
Strings / i18n backslash-in-string, colon-t, string-concat
Potential bugs dot-comment, return-error, weak-char, release-statement, public-var (.cls only)
Cross-platform run-backslash, include-case, include-backslash
Misc table-name, when-misuse
Naming naming-tt, naming-ds, naming-var, naming-param

Generative

Tool Description
gen-doc-comment Generate a formatted ABLDoc (/** */) comment block for classes, methods, functions, or procedures (powered by @breakit/abl-mcp-doc)
gen-abldoc Generate HTML documentation from existing ABLDoc comments in a project (powered by @breakit/abl-mcp-doc)
gen-ablunit-test Generate ABLUnit test class extending TestCase with ProDataSet CRUD tests

Available (disabled by default)

Tool Category Description
check-project-config Analytical Read abl.toml config
gen-business-entity Generative Generate BE .cls, Service, and Controller with ProDataSets and REST annotations
gen-workflow Generative Generate a workflow .cls with Execute + step methods, ProDataSet context
gen-business-task Generative Generate a standalone Business Task .cls with ProDataSet input/output
gen-ccs-layer Generative Generate the full CCS stack (BE + Service + Controller)
gen-openapi Generative Generate OpenAPI 3.0 spec from @openapi.openedge.export annotations
init-project Generative Scaffold a new ABL project with directory structure and abl.toml
gen-contract-tt Data Contract Generate temp-table include (.i) from schema fields
gen-contract-ds Data Contract Generate ProDataSet include (.i) wrapping the temp-table
gen-contract-json-schema Data Contract Generate JSON Schema from table/field definition
gen-contract-typescript Data Contract Generate TypeScript interface from table/field definition

Architecture

abl-mcp-server
├── config.yaml                # Default tool enable/disable + 37 lint rules
├── src/
│   ├── index.ts               # Bootstrap: auto-discovers tools, registers MCP handlers
│   ├── config-loader.ts       # Load + parse per-project YAML config (+ project overlay)
│   ├── types.ts               # ToolModule interface + config types
│   └── tools/                 # 24 pluggable tool modules (auto-discovered)
│       ├── read-abl-file.ts
│       ├── analyze-dependencies.ts
│       ├── abl-lint.ts
│       ├── find-annotations.ts
│       ├── gen-business-entity.ts
│       ├── gen-workflow.ts
│       ├── gen-business-task.ts
│       ├── gen-doc-comment.ts
│       ├── gen-abldoc.ts
│       ├── gen-contract-*.ts
│       └── ...
├── @breakit/abl-mcp-core      # Pure analysis layer — parsers, analysis, linting
├── @breakit/abl-mcp-generators # Scaffolding templates — BE, Service, Controller, Workflow
├── @breakit/abl-mcp-contracts  # Data contract generators — .i, JSON Schema, TypeScript
└── @breakit/abl-mcp-doc        # Documentation utilities — ABLDoc parser + comment generator

Installation

The server is distributed as a GitHub package (not published to npm). Install directly from the repo:

# npm
npm install github:breakit/abl-mcp-server

# pnpm
pnpm add github:breakit/abl-mcp-server

# yarn
yarn add github:breakit/abl-mcp-server

One-shot usage (no install)

npx github:breakit/abl-mcp-server

As an MCP server dependency

Add to your project's package.json:

"dependencies": {
  "@breakit/abl-mcp-server": "github:breakit/abl-mcp-server"
}

Then import in your MCP host:

import { createServer } from '@breakit/abl-mcp-server'

Development

git clone https://github.com/breakit/abl-mcp-server.git
cd abl-mcp-server
yarn install
yarn build
yarn start

Local Multi-Repo Development

If you are working on the sibling repos in ../abl-mcp-core, ../abl-mcp-contracts, ../abl-mcp-doc, and ../abl-mcp-generators, bootstrap them with Yarn and symlink them into this repo:

yarn setup:local

That command:

  • runs yarn install in each sibling repo
  • builds each sibling repo so their dist/ entrypoints exist
  • symlinks them into this repo's node_modules/@breakit/
  • includes @breakit/abl-mcp-doc in the local sibling package graph
  • links ../abl-mcp-core into ../abl-mcp-generators/node_modules/@breakit/ for local runtime resolution

When you change sibling repo code, rerun:

yarn build:local-deps
yarn link:local-deps

Acknowledgments

  • Lint rules inspired by Prolint by Jurjen Dijkstra and contributors
  • ABL parsing via tree-sitter-abl
  • Language server concepts from abl-language-server
  • Naming conventions derived from Progress ABL community standards

License

MIT

推荐服务器

Baidu Map

Baidu Map

百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。

官方
精选
JavaScript
Playwright MCP Server

Playwright MCP Server

一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。

官方
精选
TypeScript
Magic Component Platform (MCP)

Magic Component Platform (MCP)

一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。

官方
精选
本地
TypeScript
Audiense Insights MCP Server

Audiense Insights MCP Server

通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。

官方
精选
本地
TypeScript
VeyraX

VeyraX

一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。

官方
精选
本地
graphlit-mcp-server

graphlit-mcp-server

模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。

官方
精选
TypeScript
Kagi MCP Server

Kagi MCP Server

一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。

官方
精选
Python
e2b-mcp-server

e2b-mcp-server

使用 MCP 通过 e2b 运行代码。

官方
精选
Neon MCP Server

Neon MCP Server

用于与 Neon 管理 API 和数据库交互的 MCP 服务器

官方
精选
Exa MCP Server

Exa MCP Server

模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。

官方
精选