1claw MCP server
{
"mcpServers": {
"1clawai": {
"command": "npx",
"args": [
"-y",
"@1claw/mcp@latest"
],
"env": {
"ONECLAW_AGENT_API_KEY": "<oneclaw_agent_api_key>",
"ONECLAW_AGENT_ID": "<oneclaw_agent_id>",
"ONECLAW_AGENT_TOKEN": "<oneclaw_agent_token>",
"ONECLAW_VAULT_ID": "<oneclaw_vault_id>",
"ONECLAW_BASE_URL": "https://...",
"BANKR_PARTNER_KEY": "<bankr_partner_key>"
}
}
}
}{
"mcpServers": {
"1clawai": {
"command": "npx",
"args": [
"-y",
"@1claw/mcp@latest"
],
"env": {
"ONECLAW_AGENT_API_KEY": "<oneclaw_agent_api_key>",
"ONECLAW_AGENT_ID": "<oneclaw_agent_id>",
"ONECLAW_AGENT_TOKEN": "<oneclaw_agent_token>",
"ONECLAW_VAULT_ID": "<oneclaw_vault_id>",
"ONECLAW_BASE_URL": "https://...",
"BANKR_PARTNER_KEY": "<bankr_partner_key>"
}
}
}
}{
"mcpServers": {
"1clawai": {
"command": "npx",
"args": [
"-y",
"@1claw/mcp@latest"
],
"env": {
"ONECLAW_AGENT_API_KEY": "<oneclaw_agent_api_key>",
"ONECLAW_AGENT_ID": "<oneclaw_agent_id>",
"ONECLAW_AGENT_TOKEN": "<oneclaw_agent_token>",
"ONECLAW_VAULT_ID": "<oneclaw_vault_id>",
"ONECLAW_BASE_URL": "https://...",
"BANKR_PARTNER_KEY": "<bankr_partner_key>"
}
}
}
}{
"servers": {
"1clawai": {
"command": "npx",
"args": [
"-y",
"@1claw/mcp@latest"
],
"env": {
"ONECLAW_AGENT_API_KEY": "${input:oneclaw_agent_api_key}",
"ONECLAW_AGENT_ID": "${input:oneclaw_agent_id}",
"ONECLAW_AGENT_TOKEN": "${input:oneclaw_agent_token}",
"ONECLAW_VAULT_ID": "${input:oneclaw_vault_id}",
"ONECLAW_BASE_URL": "${input:oneclaw_base_url}",
"BANKR_PARTNER_KEY": "${input:bankr_partner_key}"
}
}
},
"inputs": [
{
"id": "oneclaw_agent_api_key",
"type": "promptString",
"description": "Oneclaw agent api key",
"password": true
},
{
"id": "oneclaw_agent_id",
"type": "promptString",
"description": "Oneclaw agent id",
"password": true
},
{
"id": "oneclaw_agent_token",
"type": "promptString",
"description": "Oneclaw agent token",
"password": true
},
{
"id": "oneclaw_vault_id",
"type": "promptString",
"description": "Oneclaw vault id",
"password": true
},
{
"id": "oneclaw_base_url",
"type": "promptString",
"description": "Oneclaw base url",
"password": true
},
{
"id": "bankr_partner_key",
"type": "promptString",
"description": "Bankr partner key",
"password": true
}
]
}{
"mcpServers": {
"1clawai": {
"command": "npx",
"args": [
"-y",
"@1claw/mcp@latest"
],
"env": {
"ONECLAW_AGENT_API_KEY": "<oneclaw_agent_api_key>",
"ONECLAW_AGENT_ID": "<oneclaw_agent_id>",
"ONECLAW_AGENT_TOKEN": "<oneclaw_agent_token>",
"ONECLAW_VAULT_ID": "<oneclaw_vault_id>",
"ONECLAW_BASE_URL": "https://...",
"BANKR_PARTNER_KEY": "<bankr_partner_key>"
}
}
}
}file .mcp.json (in your project root)file claude_desktop_config.jsonfile ~/.cursor/mcp.jsonfile .vscode/mcp.jsonfile ~/.codeium/windsurf/mcp_config.jsoninstall npx -y @1claw/mcpversion 0.56.3
What it does
An MCP server that connects to 1claw's HSM-backed vault to fetch secrets, signing keys, and execution policies for AI agents at call time without caching values in context. It supports local, hosted, and daemon modes with policy-based access control, threat scanning, and graduated approval workflows for transactions.
- Fetch API keys and secrets into agent calls without storing them in prompts or config files.
- Sign transactions or typed data with policy evaluation, graduated approval thresholds, and simulation-based guardrails.
- Scan agent prompts and model outputs for injection attacks and security threats in local-only mode without vault credentials.
Tools this server exposes
| Tool | Does |
|---|---|
| resolve_env | Resolve environment variables from the vault at call time. |
| inspect_content | Scan content for prompt injection and threat indicators (works in local-only mode). |
| list_agent_accounts | List Safe accounts and execution bindings for an agent. |
| migrate_agent_to_safe | Migrate an agent's execution context to a Safe account. |
| get_guardrail_shadow_report | Retrieve policy simulation and guardrail enforcement reports. |
Before you enable it
An MCP server runs with whatever access you give it. Start with a development account rather than production, and check which tools the server exposes before you let an agent call them unattended.