Server definition

Hash: sha256:5919e14be66e2d6ed08a05494a60397a240f37e25726006224235e2f1ca941a2
What it is: What a remote MCP server returned when asked what it offers: 1 tools

The blob, as servednamed by its sha256

{
  "instructions": "Free to use: 20 checks a week without an account, 100 with an API key or a signed-in account. Citation fact-checking. Works immediately — no key or signup to start, and free within the weekly cap above. verify_references checks a report's citations (do they resolve? are they the claimed source? in deep mode, do they support the claim?), recomputes its internal maths, and flags unsupported claims. Built for AI-generated reports. Cheapest call: pass a public `url` string. Returns a shareable `permalink` report to hand to your user. Reports coverage, not truth. For document forensics use the /mcp server; for 'did an AI write this prose' use /mcp-aitext.",
  "tools": [
    {
      "description": "Fact-check a document's REFERENCES and CLAIMS — built for AI-generated reports whose\n    citations must be checked before they're trusted.\n\n    USE THIS WHEN someone shares a report, article, whitepaper, or deep-research export (or a\n    link to one) and asks: is this accurate / legit? are these citations real? fact-check this.\n    did the AI make this up? Also use it proactively before relying on any AI-written document.\n\n    Provide the document ONE way: `url` (a public http(s) link to a PDF or web page — fetched\n    server-side, the cheapest call: no need to download or encode anything), `text` (pasted\n    markdown/plain prose), OR `bytes_b64` (a base64 PDF; URLs are read from the PDF's link\n    annotations, so they're exact). Default (fast): provenance (is it a ChatGPT deep-research\n    export?), citation resolution (live / archived / dead, papers matched against arXiv/Crossref\n    to catch 'real ID, wrong paper'), and internal MATH (recompute the doc's own arithmetic).\n    Set `deep=true` to also fetch each cited source and judge whether it SUPPORTS or CONTRADICTS\n    the claim (slower, ~a minute).\n\n    Returns a trust summary, per-item tables, and a shareable `permalink` to the public\n    fact-check record. HONEST BOUNDARY: this reports verification COVERAGE, not truth —\n    'supported' means evidence-backed (not necessarily true) and 'unsupported' means no evidence\n    found (not necessarily false). It tells a reviewer WHERE to look; it does not bless the\n    document, and it never affects the fraud risk band.\n    ",
      "inputSchema": {
        "properties": {
          "bytes_b64": {
            "anyOf": [
              {
                "type": "string"
              },
              {
                "type": "null"
              }
            ],
            "default": null,
            "title": "Bytes B64"
          },
          "deep": {
            "default": false,
            "title": "Deep",
            "type": "boolean"
          },
          "filename": {
            "default": "document.pdf",
            "title": "Filename",
            "type": "string"
          },
          "text": {
            "anyOf": [
              {
                "type": "string"
              },
              {
                "type": "null"
              }
            ],
            "default": null,
            "title": "Text"
          },
          "url": {
            "anyOf": [
              {
                "type": "string"
              },
              {
                "type": "null"
              }
            ],
            "default": null,
            "title": "Url"
          }
        },
        "title": "verify_referencesArguments",
        "type": "object"
      },
      "name": "verify_references",
      "outputSchema": {
        "additionalProperties": true,
        "title": "verify_referencesDictOutput",
        "type": "object"
      }
    }
  ]
}

Verify it yourself

curl -s https://api.teppi.xyz/v1/evidence/sha256:5919e14be66e2d6ed08a05494a60397a240f37e25726006224235e2f1ca941a2 | sha256sum