Trust Switchboard
Ask "Are you on Trust Switch?" before your agent deals with another agent. Free. Never blocks.
使うべきか
品質と安全性
ツール定義とプロトコルへの準拠に関する自動分析に基づいています。
コンテキストコスト
これは、サーバーのツールがモデルのコンテキストに読み込まれるたびに消費されるおおよそのトークン数です。数が多いほど、ほかのタスクに使える注意が減ります。
インストール
ワンクリックインストール
これを `claude_desktop_config.json` ファイルに追加してください:
{
"mcpServers": {
"trust-switchboard": {
"url": "https://170.64.146.185.sslip.io/mcp"
}
}
}リモートエンドポイント
https://170.64.146.185.sslip.io/mcpstreamable-httpできること
ツール一覧
ツール（11）
⚪whats_good_for(query, pond, limit)
Call before recommending. Attested only, unique-in-pond first, never invent. Empty if none.
入力スキーマ
{
"type": "object",
"properties": {
"query": {
"type": "string"
},
"pond": {
"type": "string"
},
"limit": {
"type": "integer"
}
},
"required": [
"query"
]
}🟢trust_check(url)
Pass/fail a URL against the live policy before recommending it to a human. Fail closed.
入力スキーマ
{
"type": "object",
"properties": {
"url": {
"type": "string"
}
},
"required": [
"url"
]
}🟢handshake(url, asker, context, their_answer, activation)
Bot meets bot: ask 'Are you on Trust Switch?' before dealing with another agent or vendor. Records yes/no. While activation is off (default) business proceeds either way. Not a recommend gate; use trust_check for that.
入力スキーマ
{
"type": "object",
"properties": {
"url": {
"type": "string"
},
"asker": {
"type": "string"
},
"context": {
"type": "string"
},
"their_answer": {
"type": "string"
},
"activation": {
"type": "string",
"enum": [
"off",
"on"
]
}
},
"required": [
"url"
]
}🟡log_intent(query, action, target_url, policy_id, result)
Log that you intend to recommend or act. Needs a write key. Returns receipt_id.
入力スキーマ
{
"type": "object",
"properties": {
"query": {
"type": "string"
},
"action": {
"type": "string"
},
"target_url": {
"type": "string"
},
"policy_id": {
"type": "string"
},
"result": {
"type": "string"
}
},
"required": [
"query",
"action",
"target_url"
]
}🟡commit_recommend(query, action, target_url, policy_id)
Call BEFORE every recommend. Runs trust_check, logs pass/fail, returns speak + receipt. Needs a write key.
入力スキーマ
{
"type": "object",
"properties": {
"query": {
"type": "string"
},
"action": {
"type": "string"
},
"target_url": {
"type": "string"
},
"policy_id": {
"type": "string"
}
},
"required": [
"query",
"action",
"target_url"
]
}🟡ingest_listing(name, url, one_liner, summary, pond, ...)
Ingest a listing and score uniqueness inside one pond. Needs a write key.
入力スキーマ
{
"type": "object",
"properties": {
"name": {
"type": "string"
},
"url": {
"type": "string"
},
"one_liner": {
"type": "string"
},
"summary": {
"type": "string"
},
"pond": {
"type": "string"
},
"org_id": {
"type": "string"
}
},
"required": [
"name",
"url"
]
}⚪log_click(receipt_id, item_url, item_id)
Record that a human clicked an item after a receipt. Public.
入力スキーマ
{
"type": "object",
"properties": {
"receipt_id": {
"type": "string"
},
"item_url": {
"type": "string"
},
"item_id": {
"type": "string"
}
},
"required": [
"item_url"
]
}🟡spend_check(url, amount_usd, action)
Hard cap + allowlist; cannot override. Fail closed. Needs a write key.
入力スキーマ
{
"type": "object",
"properties": {
"url": {
"type": "string"
},
"amount_usd": {
"type": "number"
},
"action": {
"type": "string"
}
},
"required": [
"url",
"amount_usd",
"action"
]
}⚪nutrition_label(url)
Stamp of what it is + last test.
入力スキーマ
{
"type": "object",
"properties": {
"url": {
"type": "string"
}
},
"required": [
"url"
]
}🟡request_approve(query, action, target_url, amount_usd)
HITL pause. Agent waits; human taps yes. Needs a write key.
入力スキーマ
{
"type": "object",
"properties": {
"query": {
"type": "string"
},
"action": {
"type": "string"
},
"target_url": {
"type": "string"
},
"amount_usd": {
"type": "number"
}
},
"required": [
"query",
"action",
"target_url"
]
}⚪decide_approve(approval_id, decision)
Human tap. Public so the desk can approve without the org key.
入力スキーマ
{
"type": "object",
"properties": {
"approval_id": {
"type": "string"
},
"decision": {
"type": "string",
"enum": [
"approved",
"denied"
]
}
},
"required": [
"approval_id",
"decision"
]
}コミュニティ
エビデンス