Reduce manual review time for AI-generated outputs.
MCP + Glama
Source verification tools for AI agents
Ground Truth helps AI workflows check claims, citations, and sources before relying on generated answers.
Use Ground Truth to reduce AI output review time so agents can verify claims before acting on them.
- Free endpoint checks for a low-friction first use case.
- Additional tools for pricing, competitors, claims, and hypotheses.
- Useful when AI output should be tied to live sources instead of stale training assumptions.
How this reduces risk
How this reduces risk
Lower the risk of acting on unsupported claims.
Add usage-based verification to agents, tools, and workflows.
MCP
Minimal MCP framing, no extra infrastructure story
MCP matters here because it is how an AI client discovers and calls the tools. That is the relevant part for buyers and builders evaluating the product right now.
Install — Claude Desktop / Cursor / Windsurf / Codex CLI
{
"mcpServers": {
"ground-truth": {
"url": "https://ground-truth-mcp.anishdasmail.workers.dev/mcp"
}
}
}Add the same URL when registering an MCP server in Cursor, Windsurf, or the Codex CLI. No API key is needed for the free first call; add an X-API-Key header for paid tools. The manifest is available at /apps/ground-truth/mcp.json.
Tools
Free tools and paid tools
Start with check_endpoint. No signup or API key. Authenticate with X-API-Key only for paid tools or after free limits.
Free
check_endpoint, inspect_security_headers, list_resources, and verify_claim (up to 5 calls/month). No API key.
Paid checks
check_pricing, compare_pricing_pages, compare_competitors, estimate_market, assess_compliance_posture, and test_hypothesis.
Monitors (paid)
Saved monitors with scheduled re-checks, evidence history, and change reports: create_monitor, run_monitor_now, get_monitor_result, list_monitors, delete_monitor.
Pricing
Plans
First endpoint/security checks and verify_claim (5/month). No signup.
10 saved monitors, weekly checks, and evidence history.
Daily checks, reports, and higher request quota.
Webhooks, alert routing, and higher monitor volume.
Endpoint checking makes the product easy to try without a long onboarding story.
The current product shape spans endpoint, pricing, market, competitor, claim, and hypothesis workflows.
Connect it through a client that already supports MCP instead of building custom glue first.
FAQ
Questions teams ask when evaluating an MCP verification layer
What is Ground Truth?
Ground Truth is an MCP server that gives AI workflows access to live verification tools before the agent turns a guess into an answer.
What does MCP mean here?
It means the tool is exposed through Model Context Protocol so compatible clients can discover and call it directly during an AI workflow.
Why would a team add this to an agent workflow?
Because pricing, endpoints, competitors, and policy language change. Verification helps the workflow check first instead of relying on memory alone.
What is the first thing to try?
Start with endpoint checking. It is the simplest way to see the product value without reworking the rest of the stack.
Can I self-host it?
Yes. The project is open source, while the hosted version keeps the trial path short for teams that want to move faster.