iamf-sentinel-mcp

iamf-sentinel-mcp

MCP server exposing IAMF conformance validation and packaging tools, enabling agents to validate IAMF files and compile/manage packaging plans through natural language, with read-only operation by default and optional execution.

Category
访问服务器

README

iamf-sentinel-mcp

An MCP server exposing iamf-sentinel (IAMF conformance validator) and iamf-loom (IAMF packager) to MCP clients — Claude Desktop, Claude Code, and any other agent runtime that speaks the protocol.

IAMF is the Alliance for Open Media's Immersive Audio Model and Formats (v1.1.0). This server lets an agent validate IAMF files against the spec's conformance rules and compile packaging manifests into inspectable, deterministic plans — without executing anything (execution is a separate, per-launch opt-in — see below).

Code comments throughout cite an internal numbered design docset (doc NN), ADRs, and pre-registered expectation labels — DESIGN-NOTES.md explains the notation and indexes every cited document.

Tools

Tool What it does Needs toolchain
iamf_validate Full conformance report on a .iamf / IAMF-in-MP4 file: S-code findings with severities, summary, stream facts. strict promotes WARN to failing. no
iamf_inspect Structure orientation: profiles, codec configs (incl. Opus pre-skip), element/presentation/frame counts, OBU trim totals, MP4 edts/elst presence. no
loom_compile Compile a Loom manifest (YAML/JSON) to its packaging plan — validate-only. Per-target backend/muxer routing with rationale, or M-code diagnostics. no
loom_explain loom explain as text: the compiled plan as its own justification. no
loom_run Compile and execute a manifest: encode, mux, measure, and gate every output through the Sentinel validator. Returns per-target outputs (sha256), gate verdicts, measured loudness, and the run-ledger path. Only registered when the server is launched with --enable-run. yes

Resources

  • checks://catalog — every S-code check the validator can emit: severity, layer, title, description, related F-register entries.
  • mcodes://catalog — every M-code compile diagnostic the packager can report, with its stable summary (retired codes kept for the record).
  • fregister://catalog — the failure-mode register (F_TO_CHECK.md): the WP1/WP3 failure catalogue mapped to Sentinel checks, as markdown.

Verified platforms. Every push runs this server's test suite on Linux, macOS, and Windows against Python 3.11 and 3.12 — ci.yml is the claim; the matrix is the evidence. Two boundaries come with it. First, this package is a thin wrapper: a green matrix here says the server behaves on those platforms, not that the conformance logic does — that is gated in iamf-sentinel and iamf-loom by their own matrices. Second, no encoder toolchain is installed in CI, so loom_run's execution path is exercised only through its failure behaviour — that a missing binary yields an actionable error rather than a traceback. Its subprocess half is unverified here by design; iamf-loom owns that claim. The matrix additionally carries a wheel-only leg with no sibling source checkout, which is what pip install iamf-sentinel-mcp actually gives you. Sample-gated tests skip in that environment by design. Nothing here is claimed for a platform that does not have a green leg.

Run

pip install iamf-sentinel-mcp
iamf-sentinel-mcp                # stdio transport, read-only (default)
iamf-sentinel-mcp --enable-run   # additionally registers loom_run (executes)

Claude Desktop / Code config:

{ "mcpServers": { "iamf-sentinel-mcp": { "command": "iamf-sentinel-mcp" } } }

Scope and posture

  • Read-only by default. Without --enable-run, tools parse and compile; nothing is executed, no files are written, no network is touched.
  • Execution is opt-in, per launch. --enable-run registers loom_run, which runs the encoder toolchain as subprocesses and writes files; the Sentinel gate inside loom run stays on by default. A missing toolchain is an actionable error naming the expected binary path. Sentinel L3 rendered-loudness QC remains unexposed.
  • The server is a thin wrapper over the products' public APIs; conformance logic lives in the products, behind their own test gates.

Apache-2.0 (see LICENSE / NOTICE), like everything else in the iamf-sentinel / iamf-loom stack. Support posture: see the core repo's SUPPORT.md.

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