Skip to main content
Glama
oathe-mcp
by oathe-ai

oathe-mcp

The Oathe Scan MCP. Oathe Scan is the audit engine of the Oathe runtime: check any skill before your agent installs it.

The runtime is a separate package: npm install -g @oathe/oathe (oathe-ai/oathe).

Quick Start

npx oathe-mcp

No API key required. No configuration needed.

Related MCP server: MCP Shield

MCP Client Configuration

Claude Desktop

Add to your claude_desktop_config.json:

{
  "mcpServers": {
    "oathe": {
      "command": "npx",
      "args": ["-y", "oathe-mcp"]
    }
  }
}

Claude Code

claude mcp add oathe -- npx -y oathe-mcp

Tools

submit_audit

Submit a GitHub or ClawHub URL for a security audit.

{ "skill_url": "https://github.com/owner/repo" }

Returns audit_id to track progress. Rate limited: one submission per 60 seconds per IP.

check_audit_status

Check the status of a submitted audit.

{ "audit_id": "uuid-from-submit" }

Poll every 5 seconds. Terminal statuses: complete, failed.

get_audit_report

Get the full security audit report for a repository.

{ "owner": "anthropics", "repo": "claude-code" }

Returns trust score, verdict, findings, category scores, and recommendation.

get_skill_summary

Get a lightweight summary (score + verdict) without full findings.

{ "owner": "anthropics", "repo": "claude-code" }

Returns score, verdict, recommendation, and finding counts.

search_audits

Search completed audits by verdict or minimum trust score.

{ "verdict": "SAFE", "min_score": 80, "sort": "trust_score", "order": "DESC" }

Returns up to 100 results.

Configuration

Environment Variable

Default

Description

OATHE_API_BASE

https://audit-engine.oathe.ai

Override the API base URL

Setting an invalid OATHE_API_BASE will produce a clear error at startup.

License

MIT

Available Tools

5 tools
check_audit_statusA

Check the status of an Oathe security audit submitted via submit_audit. Wait 90 seconds after submission before first poll, then poll every 10 seconds until status is "complete" or "failed". Statuses: queued, scanning, analyzing, summarizing, finalizing, complete, failed. Terminal statuses: complete, failed. When complete, the response includes the full audit report with trust score, verdict, and findings.

Inputs: audit_id (required).

get_audit_reportA

Get the full behavioral security audit report for a GitHub repository. Use this to review all findings before installing a third-party MCP server, plugin, or tool. Returns the latest completed audit with trust score, verdict, findings, category scores, and recommendation. Use get_skill_summary for a quick safety check instead.

Inputs: repo (required), owner (required).

get_skill_summaryA

Check if a GitHub repository is safe to install as an MCP server or AI agent skill. Returns trust score, verdict, and recommendation from Oathe's behavioral security audit. Use this BEFORE installing any third-party tool — it's the quickest safety check. For the full report with all findings, use get_audit_report instead.

Inputs: repo (required), owner (required).

search_auditsA

Search all completed Oathe behavioral security audits. Find which skills have been audited, filter by verdict or minimum trust score. Returns up to 100 completed audits. Use this to check if a skill has already been audited before submitting a new audit.

Inputs: sort, order, verdict, min_score.

submit_auditA

Submit a third-party skill for a behavioral security audit before installing it. Accepts any GitHub or ClawHub URL. Returns an audit_id to track progress. Rate limited: one submission per 60 seconds per IP. Returns existing audit_id if URL was already scanned (deduplicated: true). Use check_audit_status to poll for results.

Inputs: skill_url (required), force_rescan, notification_email.

Tool Schema Changelog

Recent tool additions, removals, and schema changes observed during successful MCP inspections.

Loading tool schema changelog…

    TDQS

    A4.4/5.0

    Scored across 5 tools

    Disambiguation5/5

    Each tool targets a distinct action: submitting, checking status, retrieving a fast summary, retrieving a full report, and searching. Even the two report-style tools are clearly separated by their descriptions, which explicitly tell when to use each.

    Naming Consistency5/5

    All tool names follow a consistent verb_noun pattern in snake_case: submit_audit, check_audit_status, get_audit_report, get_skill_summary, search_audits. The pattern is uniform and easy to predict.

    Tool Count5/5

    The 5 tools are well-scoped for a focused security-audit service: submission, status tracking, report retrieval, quick check, and historical search. No tool feels redundant or missing.

    Completeness5/5

    The tool surface fully covers the audit lifecycle: submit an audit, poll for status, retrieve the report or a quick summary, and search past audits. There are no obvious dead ends or critical gaps.

    Maintenance

    ActivityMaintained
    ResponsivenessNo recent community issues

    Related MCP Connectors

    Related MCP Servers

    • A
      license
      C
      quality
      D
      maintenance
      Security scanner and MCP server that catches dangerous patterns in MCP servers and AI agent projects, such as leaked secrets, shell execution, and prompt-injection text. Runs as both a CLI and MCP server with CI-friendly severity gates.
      2
      1
      MIT
    • A
      license
      Not graded
      quality
      B
      maintenance
      Audits MCP server configurations for security risks including capability inventory, SSRF, prompt injection, and drift detection. Works in read-only mode and can also be used as an MCP server to let AI agents audit their own attack surface.
      182 PyPI
      4
      MIT