広告掲載
MCP.so
ログイン

Repo Test Architect MCP Server認証済み注目

@RepoAssay

リポジトリホームページドキュメント

基本情報

カテゴリ

開発者ツール

ライセンス

MIT

作者

RepoAssay

投稿者

Mikael Stenberg

Repo Test Architect MCP Server について

Local-first, read-only MCP server for evidence-backed repository test architecture audits and test planning.

#testing#test-strategy#repository-analysis#static-analysis#code-quality#local-first

Repo Test Architect MCP Server の設定

以下の設定を使って、このサーバーを MCP 対応クライアントに追加してください。

{
  "mcpServers": {
    "repo-test-architect": {
      "command": "npx",
      "args": [
        "--yes",
        "repo-test-architect",
        "mcp"
      ]
    }
  }
}

Repo Test Architect MCP Server のツール

19

Start here for an unfamiliar repository or a general test-architecture review. Detect and audit all project roots once, then return the complete deterministic summary, blockers, findings, ranking, plan, execution hints, verification commands, and stats. Prefer this over audit_repo unless one project root and adapter were explicitly selected.

List registered language adapters available to audit repositories.

List deterministic project marker rules and ignored directories used during project detection.

Discover project roots, support status, and matching adapters without auditing source targets. Use this when project boundaries are needed before a specialized workflow; use analyze_repository for a complete review. Reads local repository markers and returns project-detection/v1 without executing tests or writing files.

Return raw per-project audit artifacts for all detected supported roots and report unsupported roots. Use when a downstream specialist tool needs project-audits/v1; for a complete first-pass review, prefer analyze_repository.

Summarize an existing project-audits/v1 artifact. Use only when a compact coverage view is needed separately; analyze_repository already includes this result.

Rank candidates from an existing project-audits/v1 artifact while preserving project identity. analyze_repository already includes this result.

Generate a project-aware plan from an existing project-audits/v1 artifact. analyze_repository already includes this result.

Collect concise top findings from an existing project-audits/v1 artifact. analyze_repository already includes this result.

Derive advisory test-placement findings across an existing project-audits/v1 artifact while preserving project identity. Use this for multiple audited projects; use analyze_test_placement for one audit/v1 artifact. Does not rescan the repository, execute tests, or create, move, or modify files.

Aggregate coverage, counts, risk and signal distributions, frameworks, and adapter usage from an existing project-audits/v1 artifact. Use this for separate reporting or comparisons; analyze_repository already includes the same stats. Returns project-stats/v1 without rescanning the repository, executing tests, or writing files.

Audit one explicitly selected project root and return audit/v1. Use only when the caller selected a single adapter or project boundary; adapterId defaults to javascript. For an unfamiliar or complete repository review, use analyze_repository.

Validate and return an existing audit/v1 artifact unchanged; this does not scan a repository. Use audit_repo or analyze_repository for repository discovery.

Create a deterministic plan/v1 from an existing single-project audit/v1 artifact. Use this after audit_repo, optionally selecting one exact plan item; use generate_project_test_plan for multiple audited projects. Does not rescan the repository, execute tests, or generate or write test code.

Derive provider-neutral execution, context, parallel-safety, and repository-reasoning hints from a plan without invoking models or subagents.

Explain the evidence, risk, testability, and recommendation for one target in an existing audit/v1 artifact. Use this after audit_repo or audit_projects when one target needs detail beyond the audit summary. Returns target-explanation/v1 without rescanning the repository, executing tests, or writing files.

Rank testable audit targets by risk reduction and maintenance cost.

Derive advisory test-placement findings from an existing single-project audit/v1 artifact. Use this for one audited project; use analyze_project_test_placement for a project-audits/v1 artifact. Does not rescan the repository, execute tests, or create, move, or modify files.

Return a structured deferred result. This tool does not generate or write test code while native generation remains disabled.

概要

• ## What is Repo Test Architect?

Repo Test Architect is a local-first MCP server for evidence-backed repository test audits and planning.

It detects project roots, maps source-to-test evidence, identifies blockers and weak coverage, and produces deterministic findings, ranked test candidates, test plans, and repository-native verification commands.

Supported ecosystems

Ten bounded, supported adapters cover JavaScript/TypeScript, C#/.NET, Elixir, Go, Kotlin/JVM, PHP, Python, Ruby, Rust, and Swift.

Version 1.0.0-beta.2 adds experimental Dart and Flutter support. Dart evidence is structural: imports and exports do not prove calls, assertions, or successful test execution.

What's new in beta.2?

  • Experimental Dart and Flutter auditing.
  • More conservative evidence handling for comments, strings, and heredocs.
  • Stronger symlink and nested-package ownership boundaries.
  • More accurate PHPUnit test discovery and Kotlin/Swift verification-command selection.
  • Security dependency updates.

The existing CLI contracts, 19 MCP tools, and versioned artifact schemas remain compatible.

Installation

Node.js 20 or newer is required.

To opt into the public beta, add this configuration to your MCP-compatible client:

{
  "mcpServers": {
    "repo-test-architect": {
      "command": "npx",
      "args": ["--yes", "repo-test-architect@beta", "mcp"]
    }
  }
}

The @beta channel currently selects 1.0.0-beta.2. For a fixed version, replace repo-test-architect@beta with repo-test-architect@1.0.0-beta.2.

The unqualified package name continues to install the public alpha, 0.3.0.

For a general repository review, connected models should begin with analyze_repository.

Privacy and limitations

Repository analysis runs locally. The MCP tools are read-only; source remains on the machine unless the MCP host or another configured tool sends it elsewhere.

Findings are evidence-backed review input, not proof of complete test coverage or instructions to change code automatically. Each adapter supports documented, bounded repository shapes.

Native test generation remains intentionally deferred. This is a public beta, not an RC or stable release.

Beta.2 release notes · Documentation

コメント

「開発者ツール」の他のコンテンツ

LocalCan

LocalCan

Gives AI agents public URLs (tunnels) for localhost, live HTTP traffic inspection, snapshot publishing, and access control. Part of LocalCan, the ngrok alternative for Mac, Windows and Linux. Free plan.

82

UXMachine

UXMachine

UXMachine measures a web page in a real browser and returns verifiable observations, each with its evidence and its limits. It opens the address on desktop (1280×800) and mobile (390×844) and follows the site's own links

how.is — domain & website intelligence

how.is

Remote MCP server for domain and website checks: DNS, WHOIS/RDAP, domain age, TLS, DNSSEC, email auth (SPF/DKIM/DMARC), blocklists, a trust verdict ("should my agent click, sign in to, or pay this site?"), availability +

Nebelus

Nebelus

Nebelus is an enterprise agentic-AI platform for regulated industries. This MCP server exposes the Nebelus Construction API: ~48 tools that give an MCP client full agent build parity — create/update/probe agents, edit ag

Capgo MCP

Cap-go

Console, Backend and CLI to manage Capgo Instant update and Native build for Capacitor apps Build and upload to app store your mobile app

広告掲載