Drop-in recipes to give any AI agent one capability: verify a photo's capture time and provenance (cryptographic C2PA Content Credentials validation, EXIF and XMP metadata, and classical pixel forensics) and return one verdict with a confidence and a signed audit record.
Provenance-first, not a deepfake or AI-generation detector. Results are investigative triage, not proof.
- Website: https://chronoverify.com
- Integrations: https://chronoverify.com/integrations
- API and method: https://chronoverify.com/method#api
- npm (MCP server): https://www.npmjs.com/package/chronoverify-mcp
- npm / PyPI (SDKs):
chronoverify
Or paste this into your MCP client config (Claude Desktop, Claude Code, Cline, Windsurf):
{
"mcpServers": {
"chronoverify": {
"command": "npx",
"args": [
"-y",
"chronoverify-mcp"
],
"env": {
"CHRONOVERIFY_API_KEY": "cv_live_..."
}
}
}
}Omit the env block to use the free, rate-limited public path. See
mcp/install.md for every client.
Copy-paste tool definitions:
tool-definitions/openai-chat-completions.jsontool-definitions/openai-responses.jsontool-definitions/anthropic-messages.json
When the model calls chronoverify_verify_image, POST the arguments to
https://chronoverify.com/v1/verify and return the JSON verdict.
Point each framework's MCP adapter at the published server:
Runnable examples on the first-party SDKs:
- Python (
examples/example.py):pip install chronoverify - TypeScript (
examples/example.ts):npm i chronoverify
skill/SKILL.md is a ready Claude skill that calls the MCP tool
or the REST API.
MIT