Records agent tool-call evidence, checks approval rules, summarizes risk timelines, and exports audit receipts over remote MCP.
Copy the AI prompt to install this server into Claude Code, Cursor, or another agent ā or use 1-click editor setup below.
š” Paste the JSON block into your client's configuration file under mcpServers, then restart the application.
Inspect callable tools, capabilities, and parameters exposed to AI agents by Toolcallwitness MCP.
create_witness_receiptCallable MCP tool function
check_approval_ruleCallable MCP tool function
summarize_risk_timelineCallable MCP tool function
export_audit_receiptCallable MCP tool function
The toolcallwitness-mcp MCP server connects an MCP client to the hosted ToolCall Witness service at https://toolcallwitness.clauxel.com/mcp. Its purpose is to provide a witness layer around AI agent tool calls, including receipts, approval evidence, argument redaction, result summaries, and audit-oriented exports.
Four tools are available through MCP:
create_witness_receipt creates a witness receipt for a tool call.check_approval_rule checks an approval rule.summarize_risk_timeline produces a risk timeline summary.export_audit_receipt exports an audit receipt.These capabilities fit workflows that need a record of agent actions or a way to prepare tool-call activity for review.
The service is hosted remotely rather than shipped as executable production source in this repository. An MCP client sends JSON-RPC requests to the /mcp endpoint using Streamable HTTP. Each production request must include a bearer token in the Authorization header:
The repository describes the project as a documentation and directory-submission reference for the hosted service. It does not contain the private production implementation. Browser requests without authentication return a JSON error rather than internal service details.
To use the toolcallwitness-mcp MCP server, open the product website, select a plan, and create or request an API token. Configure an MCP client that supports remote Streamable HTTP servers with the service endpoint and token, then send JSON-RPC requests.
The public endpoint is:
Authentication is configured as a request header, not as a documented server-side environment variable. The README does not provide client-specific configuration files or a local installation procedure.
The toolcallwitness-mcp MCP server exposes a compact set of tools focused on evidence and review of agent tool calls:
The available material does not define each tool's input schema, output schema, approval-rule language, or receipt format, so those details should be confirmed through the service's MCP tool definitions.
The toolcallwitness-mcp MCP server is a paid hosted service. Production access depends on a bearer token issued through the product website, and the repository does not provide a local server implementation. Client compatibility therefore depends on support for remote Streamable HTTP MCP servers and bearer-token configuration.
The README does not identify a specific license for the repository or document retention, pricing details, rate limits, or data-handling policies. Review the product site and server card before adopting it for sensitive or regulated tool-call data.
Factual signals from GitHub, npm, and our automated checks ā not a rating.
No reviews yet ā be the first to share how this listing worked for you.
Showcase your server listing on GitHub or your project documentation. Embed this dynamic SVG badge to highlight official listing status and live engagement.
[](https://allmcps.com/mcp/toolcallwitness-mcp)<a href="https://allmcps.com/mcp/toolcallwitness-mcp"><img src="https://allmcps.com/api/badge/toolcallwitness-mcp?style=directory" alt="Toolcallwitness MCP on AllMCPs" /></a>