mtl-c2pa-mcp
MCP server for reading C2PA content provenance manifests from media files, enabling natural language queries about AI-generated content, signatures, and provenance.
README
mtl-c2pa-mcp
MCP server for reading C2PA content provenance manifests from media files — built for Claude Code.
⚠️ Experimental — Early development. The C2PA spec and tooling are still evolving. Don't rely on this for legal evidence or compliance work.
C2PA (Coalition for Content Provenance and Authenticity) is the open standard behind Adobe's Content Credentials and is now embedded in Google Lyria AI-generated MP3s. Each manifest records who created the file, what tool produced it, whether it's AI-generated, what watermarks (e.g. SynthID) were applied, and who signed the claim.
This MCP wraps the official c2pa-python library so Claude can read those manifests in plain language.

Tools
| Tool | Description |
|---|---|
c2pa_summary |
Human-friendly summary: generator, AI flag, actions, ingredients, watermarks, signature |
c2pa_read |
Full raw manifest store as JSON |
c2pa_assertions |
List all assertions from the active manifest |
c2pa_ingredients |
List ingredients (sources used) from the active manifest |
c2pa_verify |
Signature issuer, validation state, successes & failures |
c2pa_scan |
Audit a directory: which files carry C2PA, which are AI-generated |
c2pa_info |
Library version and supported MIME types |
Supported formats
The underlying c2pa-python library supports many formats. Most relevant here:
- Audio: MP3, WAV, FLAC, OGG, M4A, AAC
- Image: JPG, PNG, WebP, TIFF, AVIF, HEIC, GIF, SVG
- Video: MP4, MOV, AVI
- Other: PDF
Call c2pa_info for the full live list.
Setup
git clone https://github.com/musictechlab/mtl-c2pa-mcp.git
cd mtl-c2pa-mcp
poetry install
Claude Code configuration
claude mcp add -s user mtl-c2pa -- poetry --directory /path/to/mtl-c2pa-mcp run python -m mtl_c2pa_mcp
Usage examples
Once configured, just ask Claude in natural language.
Quick summary of a Lyria-generated MP3
"What does the C2PA manifest in
~/Downloads/Sovereign_Ascent.mp3say?"
{
"file": "/Users/you/Downloads/Sovereign_Ascent.mp3",
"active_manifest": "urn:c2pa:80faaddf-fe27-1e7d-0ce5-4a70eeba2dd1",
"generator": {
"name": "Google C2PA Core Generator Library",
"version": "916434528:916944653"
},
"is_ai_generated": true,
"digital_source_types": [
"http://cv.iptc.org/newscodes/digitalsourcetype/trainedAlgorithmicMedia"
],
"actions": [
{
"action": "c2pa.created",
"digitalSourceType": "http://cv.iptc.org/newscodes/digitalsourcetype/trainedAlgorithmicMedia",
"description": "Created by Google Generative AI."
},
{
"action": "c2pa.edited",
"digitalSourceType": "http://cv.iptc.org/newscodes/digitalsourcetype/trainedAlgorithmicMedia",
"description": "Applied imperceptible SynthID watermark."
}
],
"ingredients_count": 0,
"watermarks": [
{
"label": "c2pa.actions.v2",
"action": "c2pa.edited",
"description": "Applied imperceptible SynthID watermark."
}
],
"signature_issuer": "Google LLC",
"validation": "valid"
}
Verify a signature
"Is the C2PA signature on
track.mp3valid? Who signed it?"
Audit a music folder
"Scan my
~/Musicdirectory and prepare a list of tracks with info if c2pa is included."

Or in raw JSON form:
{
"directory": "/Users/you/Downloads",
"total_files": 12,
"with_manifest": 3,
"files": [
{
"file": "/Users/you/Downloads/Sovereign_Ascent.mp3",
"has_manifest": true,
"generator": {"name": "Google C2PA Core Generator Library", "version": "..."},
"is_ai_generated": true,
"digital_source_types": ["http://cv.iptc.org/newscodes/digitalsourcetype/trainedAlgorithmicMedia"]
},
{
"file": "/Users/you/Downloads/photo.jpg",
"has_manifest": false
}
]
}
Inspect raw manifest
"Dump the full C2PA manifest store of
track.mp3."
Returns the same JSON shape produced by c2patool ./track.mp3 — useful when you need every field.
Why this exists
When Google shipped C2PA in Lyria MP3 downloads in May 2026, it became the first major DSP-adjacent player to embed AI provenance directly in music files. Inspecting those manifests usually means installing the Rust c2patool binary and reading raw JSON. This MCP lets you ask Claude "is this AI-generated?" and get a clean answer — same workflow as mtl-metadata-mcp and mtl-vhc, but for provenance instead of ID3 tags or human-cert registry lookups.
Tech stack
- Python 3.10+
- c2pa-python — Adobe's official Rust-backed Python binding
- MCP SDK (FastMCP)
- Poetry — dependency management
Contributing
See CONTRIBUTING.md.
Security
To report a vulnerability, see SECURITY.md.
License
MIT — see LICENSE.
<div align="center"> MusicTech Lab - Rockstars Developers dedicated to the Music Industry<br> <a href="https://musictechlab.io">Website</a> <span> | </span> <a href="https://linkedin.com/company/musictechlab">LinkedIn</a> <span> | </span> <a href="https://musictechlab.io/contact">Let's talk</a><br> Crafted by <a href="https://musictechlab.io">musictechlab.io</a> </div>
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。