Agentic Platform

io.github.andysalvo由 GitHub 帳號 andysalvo 發布v1.1.0更新於 Oct 24, 2026

Free MCP tools: the only MCP linter, health checks, cost estimation, and trust evaluation.

發布方自己的服務：agentic.no

已驗證Streamable HTTP可網頁執行開發者工具 AI 與機器學習

服務詳細資訊

識別碼: io.github.andysalvo/agentic-platform
傳輸: Streamable HTTP
執行環境: Web executable
信任: Verified
發布於: Sep 16, 2026

安全提示

MCP 服務會接收對話中的工具呼叫請求，並可能執行外部操作。啟用前請先了解該服務能做什麼，並只授予它所需的權限。

註冊中心已驗證此條目所屬的 GitHub 帳號 andysalvo。這並不表明該帳號就是產品方。

SourceWeft 的檢查只涵蓋條目是否完整、端點是否回應；不審查伺服器的程式碼，也不審查其工具的行為。

在 README 中新增徽章

向讀者展示該 MCP 服務已收錄於 SourceWeft，徽章連結到本頁。

[![Listed on SourceWeft](https://sourceweft.com/badge/mcp/io.github.andysalvo%2Fagentic-platform.svg)](https://sourceweft.com/mcp/io.github.andysalvo%2Fagentic-platform?utm_source=readme&utm_medium=badge)

概覽

AI 產生的概覽

託管式 MCP 伺服器，可檢查 MCP 工具定義、為代理系統提示詞做治理評分，並估算模型成本。

可以做什麼: 檢查 MCP 工具定義中的反模式、缺少欄位、描述品質與 schema 問題，並給出通過／失敗報告與修正建議
依治理與最佳實務為代理系統提示詞評分（0-100），並提供診斷與改善建議
估算並比較 Claude、GPT、Gemini 等模型的呼叫成本，提供單次呼叫、單次執行與每日明細
查閱關於代理治理、代理經濟與意圖架構的專家技能內容
適用情境: 適合在建立或審查 MCP 伺服器與代理提示詞時使用，快速做一次驗證、取得治理角度的提示詞評分，或在定案設計前比較模型成本。
執行需求: 需要在用戶端設定中加入遠端 MCP 端點；免費工具據稱不需 API 金鑰。README 另提到可自架的 Docker 映像，以及專家技能依 API 金鑰每日 10 次免費呼叫。

安裝前請注意

送出的提示詞與工具定義會傳送到第三方託管服務。README 列出透過 Stripe 購買的付費點數包及同一體系下的其他付費服務，使用前請確認哪些免費。README 中「唯一 MCP 檢查器」的說法屬於供應商自述。

安裝

在 SourceWeft 中

開啟儀表板中的 Agentic Platform，將其新增到工作區。
為需要使用其工具的對話啟用該服務。

Web executable，透過 Streamable HTTP。遠端服務在工作區中設定後即可從網頁執行環境執行。

交給代理程式安裝

把這段提示詞貼上到 Claude Code、Codex、Cursor 等代理程式，或 SourceWeft 對話中。代理程式會讀取該伺服器的安裝指南，先向你展示發布者、端點和登入方式，經你同意後再把它新增到你的客戶端。金鑰不會寫進任何命令。

閱讀 https://sourceweft.com/mcp/io.github.andysalvo%2Fagentic-platform/install.md，並按其中說明把這個 MCP 伺服器接入我的客戶端。修改任何設定之前，先告訴我它的發布者、端點和登入方式，等我確認。不要把 API 金鑰寫進命令或我會提交的檔案；問我金鑰應從哪裡取得。

檢視代理程式所依據的安裝指南

其他 MCP 客戶端

把它新增到你客戶端的 mcpServers 設定中。

{
  "mcpServers": {
    "agentic-platform": {
      "type": "http",
      "url": "https://platform.crestsystems.ai/mcp"
    }
  }
}

README

Agentic Platform

[agentic-platform MCP server]

The only MCP linter that exists. Plus free governance scoring and cost estimation for AI agents.

Validate your MCP tool definitions, score your agent's system prompt on governance best practices (0-100), and compare costs across all major models. No API key needed.

Installation

Remote MCP Server (Hosted)

Add to your MCP client configuration:

json

{  "mcpServers": {    "agentic-platform": {      "url": "https://platform.crestsystems.ai/mcp"    }  }}

{  "mcpServers": {    "agentic-platform": {      "url": "https://platform.crestsystems.ai/mcp"    }  }}

Claude Code

Docker

Works with Claude Desktop, Claude Code, VS Code, Cursor, Cline, and any MCP-compatible client.

Free Tools (No API Key Needed)

Tool	What It Does
`mcp_manifest_lint`	The only MCP linter. Validate your tool definitions for anti-patterns, missing fields, bad descriptions, and schema issues. Pass/fail report with fixes.
`agent_health_check`	Score your agent's system prompt on governance and best practices (0-100). Detailed diagnostic with specific issues and remediation.
`estimate_agent_cost`	Compare costs across Claude, GPT, Gemini, and other models. Per-call, per-run, and per-day breakdown with optimization tips.

Expert Skills (Free Tier: 10 Calls/Day)

Skill	Description
`governance`	Design agent systems that preserve human authority. 3 core invariants, Coupled Authority Phenomenon, 14 forbidden failure modes.
`agentic-economics`	Pricing models, unit economics, and revenue architecture for AI agent platforms. Market projections, cost-to-serve benchmarks (March 2026).
`intent-architecture`	Build deterministic, auditable systems. Append-only decision logs, 9 system invariants, implementation patterns, embodiment tests.

Quick Start

1. mcp_manifest_lint(tools_json='[{"name":"my_tool","description":"Does something"}]')   -> Pass/fail report with specific fixes
2. agent_health_check(system_prompt="Your agent's system prompt here")   -> Score 0-100 with detailed diagnostic
3. estimate_agent_cost(model="claude-sonnet-4", input_tokens=2000, output_tokens=1000, num_calls=10)   -> Cost comparison table across all major models

1. mcp_manifest_lint(tools_json='[{"name":"my_tool","description":"Does something"}]')   -> Pass/fail report with specific fixes
2. agent_health_check(system_prompt="Your agent's system prompt here")   -> Score 0-100 with detailed diagnostic
3. estimate_agent_cost(model="claude-sonnet-4", input_tokens=2000, output_tokens=1000, num_calls=10)   -> Cost comparison table across all major models

Pricing

Free tools: Unlimited. No API key needed.
Expert skills: 10 free calls/day per API key.
Credit packs: $5 (50 credits) or $20 (250 credits) via Stripe.

Crest Service Fleet

Service	URL	Purpose
Verify API	verify.crestsystems.ai	Conformance verification for x402 action_ref. Submit receipts, get signed verdicts with public permalinks.
Trust Check	supership.crestsystems.ai/check	Free pre-payment trust oracle for x402 agent commerce.
Conformance	crestsystems.ai/conformance	Public conformance results matrix for action-ref-v1.
x402 Audit	audit.crestsystems.ai	Smart contract audit, code scan, wallet risk profiling ($0.25-$1.00 USDC).
x402 Data	data.crestsystems.ai	Crypto market data ($0.01/call USDC on Base).

Governance

This platform operates under three core invariants:

No Silent Inference -- no inferring or reinterpreting input without explicit documentation
Auditability at the Decision Level -- every state change traceable to a decision
Explicit Authority Transfer -- all commitments require explicit authorization

These apply to the platform itself, not just its users.

Keywords

MCP server, MCP linter, MCP validator, MCP manifest lint, agent governance, agent health check, agent cost estimator, AI agent tools, Model Context Protocol, MCP tool validation, agent diagnostics, governance scoring, x402 conformance, action-ref verification, agent accountability, trust oracle, agent commerce