context-surface-analyzer
Enables deterministic analysis of a single Agent tool-catalog snapshot and comparison of two snapshots through read-only JSON-RPC tools, reporting digests, schema statistics, duplicates, collisions, budget checks, and diffs.
README
Context Surface Analyzer
A local, deterministic utility for measuring one explicit Agent tool-catalog snapshot and comparing two snapshots.
It reports canonical UTF-8 bytes and SHA-256 digests, tool/schema counts, exact duplicate schemas, hard name collisions, declared budget checks, and deltas. Optional token counts remain externally measured and fully labeled; the product does not infer tokens from bytes or decide whether a tool is useful.
Run
Requires Node.js 22 or newer and has no third-party runtime dependencies.
npm run check
node src/cli.js analyze examples/baseline.json
node src/cli.js diff examples/baseline.json examples/updated.json
npm start
The web surface opens at http://127.0.0.1:4173. Use --port 0 to select an
ephemeral port. A snapshot can be pasted or loaded from a bounded local JSON
file; the browser does not retain it after the page is closed.
Snapshot authority
Acquire each snapshot from the current provider, host, or an authorized
exporter. This product deliberately does not discover a host, start arbitrary
MCP servers, crawl files, or treat repository fixtures as the current installed
catalog. Project the exported catalog into the closed
context-surface.snapshot.v0.1 format before analysis and retain the exporter,
provider, and revision as the source of record.
MCP
Start the newline-delimited JSON-RPC stdio server:
npm run mcp
It exposes two read-only tools:
context.analyze: one snapshot to one analysis;context.diff: earlier and later snapshots to one comparison.
Both accept snapshot JSON as bounded strings, so the Agent route has no ambient filesystem authority. See the product model for the snapshot contract and exact semantics.
This portable runtime intentionally serves the MCP 2025 legacy era through
revision 2025-11-25. Dual-era clients can identify that boundary from the
server/discover method-not-found response and fall back to initialize.
Modern-only 2026 transport is not claimed by this dependency-free plugin.
Local Codex plugin
Build and validate the repository-local plugin, add this repository as a local marketplace, then install the plugin:
npm run check
codex plugin marketplace add <repository-root>
codex plugin add context-surface-analyzer@context-surface-analyzer
Start a new Codex task after installation so the Skill and MCP tools are loaded from the same cached plugin version.
Status
This is a local MVP. Development checks, direct runtime probes, installed Codex routing, production load behavior, public-release readiness, and owner visual acceptance remain separate claims. See the review contract and the first real-catalog dogfood note.
License
Apache License 2.0. See LICENSE and NOTICE.
推荐服务器
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。
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。
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。
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。
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。
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。
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。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。
Neon MCP Server
用于与 Neon 管理 API 和数据库交互的 MCP 服务器
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。