Dechonet MCP

Report a Page

report_page

Submit a page to DechoNet for phishing, fake payment windows, hidden AI instructions, malicious downloads, incorrect verdicts, malicious MCP endpoints, or network anomalies so it can be checked later.

Instructions

Report a page to DechoNet: phishing or brand impersonation, a fake payment window (card fields where a payment provider should be), hidden instructions aimed at AI agents, a malicious download, a DechoNet verdict that looks wrong (false_positive) or a danger it missed (missed), an MCP server whose tool descriptions steer agents (malicious_mcp — url is the MCP endpoint), or a network anomaly (network_anomaly — the certificate serial/issuer or addresses you saw for the host, in observed). Kept for 90 days: the host and the kind, with an anonymous reporter hash used to count independent reports; for malicious_mcp the endpoint path, for network_anomaly the observed values and the country of the connection. No query, page text or free text. This call only records the report and returns at once; it opens nothing. A report changes no verdict by itself: later, DechoNet's own hourly check visits the reported host (an outside site) with the same rules as phishing_link_check (mcp_server_check for malicious_mcp, dns_lookup and ssl_check for network_anomaly), and only what that check observes counts; hosts reported by several independent sources are checked first. Not read-only (it records the report); rate-limited.

Tool call template

Replace each placeholder with a value matching the input schema. Only required parameters are included; add optional parameters as needed.

{
  "id": 1,
  "jsonrpc": "2.0",
  "method": "tools/call",
  "params": {
    "arguments": {
      "url": "<url>",
      "kind": "<kind>"
    },
    "name": "report_page"
  }
}

Input Schema

Name	Required	Description
`url`	Yes	Address of the page (only its host is kept)
`kind`	Yes	What is wrong with the page
`observed`	No	network_anomaly only: the certificate serial (hex) and issuer, and the addresses, you saw for the host

Output Schema

Name	Required	Description
`host`	Yes
`kind`	Yes
`receipts`	No	Signed observation receipts behind this result, one per lookup: when, from where and under which rules version DechoNet observed it. Each url shows the record and checks its Ed25519 signature.
`received`	Yes

Schema Changelog

Changes observed during successful MCP inspections.

Addedv1.5.14Oct 15, 2026

TDQS

A4.5/5.0

Behavior5/5

Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?

Goes well past the annotations: it discloses the 90-day retention scope (host and kind only, anonymous reporter hash, endpoint path for malicious_mcp), states what is deliberately NOT stored (no query, page text or free text), confirms the call returns immediately and opens nothing, and flags that it is rate-limited. Annotations already say readOnlyHint=false, but the description adds the concrete side effects and privacy boundary an agent needs before writing.

Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.

Conciseness3/5

Is the description appropriately sized, front-loaded, and free of redundancy?

The opening sentence is a single ~90-word colon-and-em-dash chain of eight report kinds, which is hard to parse in one pass. The most decision-relevant constraints — 'only records the report and returns at once' and 'changes no verdict by itself' — sit in the middle of the block rather than being front-loaded after the purpose.

Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.

Completeness5/5

Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?

For a 3-parameter write tool with full schema coverage, annotations and an output schema, nothing essential is missing: retention, evidence requirements per kind, rate limiting, non-idempotency implications and the downstream verification path are all covered. The agent knows both what will happen and what will not.

Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.

Parameters4/5

Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?

Schema coverage is 100%, so baseline is 3, but the description adds real meaning beyond the terse schema blurbs — each enum value is glossed (fake payment = card fields where a provider should be, false_positive = a DechoNet verdict that looks wrong, missed = a danger it missed), and 'only its host is kept' clarifies what url does. The observed payload fields (serial/issuer/ips) are only implied via 'the observed' rather than spelled out.

Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.

Purpose5/5

Does the description clearly state what the tool does and how it differs from similar tools?

States a specific verb and resource ('Report a page to DechoNet') and then enumerates the exact reportable conditions, including the two special shapes (malicious_mcp where url is the MCP endpoint, network_anomaly where evidence goes in observed). This is enough for an agent to separate it from phishing_link_check, mcp_server_check, ssl_check and dns_lookup, which are named as the things that actually verify.

Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.

Usage Guidelines4/5

Does the description explain when to use this tool, when not to, or what alternatives exist?

Clearly conveys when to use it (you observed one of the listed harms) and, crucially, sets expectations that this is only a tip-off — 'A report changes no verdict by itself'. What it does not say explicitly is the inverse routing rule, e.g. 'use phishing_link_check instead if you need an immediate verdict right now'; the agent must infer that from the description of the hourly re-check.

Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.