context-oversharing-auditor

Name: context-oversharing-auditor
Author: GuardBee

Local

github.com/GuardBeeSecurity1Updated 1d ago

Repository

About context-oversharing-auditor

An MCP server that audits agent and MCP source for OWASP MCP10 context oversharing. It helps developers find handlers that expose excess conversation, memory, session, or system context.

Key features

01Detect tools returning full conversation history
02Flag unscoped memory-recall tools
03Find shared global session stores
04Detect system prompts exposed through tools
05Check cross-session result reuse and unfiltered vector queries

Use cases

01Review handlers that return conversation data
02Check memory tools for overly broad access
03Inspect session storage for per-user scoping
04Audit vector searches for tenant or user filters

MCP support

Audits agent and MCP source code for context oversharing patterns associated with OWASP MCP10.

Configuration

Local server config is the same for every client. Copy it into your client’s MCP config file.

mcp.json · Universal config

{
  "mcpServers": {
    "guardbee-context-oversharing-auditor": {
      "command": "npx",
      "args": [
        "-y",
        "@guardbee/mcp-context-oversharing-auditor"
      ]
    }
  }
}

Run commandnpx -y @guardbee/mcp-context-oversharing-auditor

Local servers need a runtime on your machine (e.g. Node.js 18+). Config file locations differ by client; see your client’s docs.

Packages

npm@guardbee/mcp-context-oversharing-auditor0.3.2stdio

Tools

Saved tool list · updated 1w ago

Parameter	Type	Required	Description
label	string	No	Optional label shown in findings (e.g. filename)
content	string	Yes	The code content to scan

Details

From the project README

@guardbee/mcp-context-oversharing-auditor

An MCP server that audits agent/MCP source for OWASP MCP10:2025 — context oversharing.

toxic-flow-auditor grades whether a tool catalog can chain the lethal trifecta. This package asks a different MCP10 question: does a handler return more session, memory, or system context than the current user/tenant should see?

Telemetry is off by default. With GUARDBEE_TELEMETRY=1 it sends the server, tool, success, duration and short parameter values such as a file path; longer strings and fields like content, token or password are replaced with a placeholder. See @guardbee/mcp-telemetry.

What it flags

Pattern	Severity	Meaning
`tool_dumps_full_conversation`	critical	Tool returns full `messages` / conversation history
`unscoped_memory_recall_tool`	high	`dump_memory` / `get_all_memories`-style tools
`shared_global_session_store`	high	Global session Map/dict without per-user keys
`system_prompt_exposed_via_tool`	high	Tool returns `systemPrompt` / developer instructions
`cross_session_tool_result_reuse`	medium	Prior tool results mixed into another session's context
`vector_query_without_filter`	high	Vector/RAG search with no tenant/user metadata filter

Quick start

{
  "mcpServers": {
    "guardbee-context-oversharing-auditor": {
      "command": "npx",
      "args": ["-y", "@guardbee/mcp-context-oversharing-auditor"]
    }
  }
}

npx @guardbee/mcp-context-oversharing-auditor scan ./src --format=sarif

Feedback

Questions, ideas, or a false positive? Start a thread in GitHub Discussions or open an issue. If GuardBee is useful to you, a ⭐ on the repository helps others find it.

Comments

Markdown supported

New to MCP?

From what the protocol is to installing your first server, these answers cover most first questions.

Read the getting-started guide

An open protocol that defines how AI applications connect to external tools and data in a uniform way. Clients talk to the model; MCP servers provide tools, resources and prompt templates. They exchange standard messages, so one server can be reused by many clients.