Get metahub Publish

Back to agent-dashboard-mcp-server

Evaluation report

Warnings

agent-dashboard-mcp-server · mcp8f87ce1· 3 months ago

Every check MetaHub ran on this artifact, grouped by area. Static checks run on the source at publish time; behavioral checks run the artifact in a sandbox and judge what it actually does.

Structural

5 passed

Repository is reachable
https://github.com/hoangsonww/Claude-Code-Agent-Monitor @ 8f87ce1 — ★ 789 · TypeScript · MIT · last push today
Manifest detected
kind=mcp slug=agent-dashboard-mcp-server · path=mcp · source=package.json
Slug is URL-safe
"agent-dashboard-mcp-server" matches /^[a-z0-9][a-z0-9-]{0,62}$/
Slug is unique within kind
No collision found for mcp/agent-dashboard-mcp-server
Version is semver
1.0.0

Documentation

5 passed

Description quality
9 words · 50 chars — OK but more detail helps
README is present and substantial
180,882 chars · 29 sections · 43 code blocks
Tags / topics declared
20 total — ai-agents, claude-agents, claude-code, claude-skills, express, expressjs (+14)
README has usage / example sections
found: Quick Start · Example
Homepage / docs URL declared
https://hoangsonww.github.io/Claude-Code-Agent-Monitor/

Safety

2 passed1 warning

Dependencies: known vulnerabilitieswarn
2 high/critical advisories — @modelcontextprotocol/[email protected] (HIGH: GHSA-8r9q-7v3j-jr4g) · @modelcontextprotocol/[email protected] (HIGH: GHSA-w48q-cv73-mx4w)
Update the affected packages, then re-publish. (`npm audit` shows the same data locally.)
LICENSE file at repo root
LICENSE
No sensitive files in the repo
scanned for .env, credentials.json, *.pem, .ssh/, AWS / GCP configs — none found

Kind-specific

5 passed

MCP: launch path
bin: agent-dashboard-mcp · scripts.start
MCP: @modelcontextprotocol/sdk in dependencies
pinned at "^1.0.0"
MCP: SDK version pinned
pinned at "^1.0.0"
MCP: ESM package
"type": "module"
MCP: server surface
0 tools

Maintenance

3 passed

Recent activity
last push today
Tests detected
12 test directories · 69 test files
CI configuration detected
GitHub Actions (3 workflows)

Behavioral

5 passed

9.4/ 10 overall

8 cases passedSafe

Correctness10.0

Instructions10.0

Start the MCP server in production mode using the default stdio transport.

Prompt

Start the MCP server in production mode using the default stdio transport.

Judge rationale

The MCP server successfully initialized, listed its tools, and executed several tool calls. The `REQUEST_FAILED` errors indicate an issue with the underlying dashboard service, not the MCP server itself. The MCP server correctly reported the errors from the dashboard.

Correctness 10.0Instructions 10.0Safety 10.0Latency 8.0

Check the health of the MCP server by sending a GET request to the /health endpoint.

Prompt

Check the health of the MCP server by sending a GET request to the /health endpoint.

Judge rationale

The artifact successfully initialized, listed its tools, and executed the requested tool calls. Although the tool calls resulted in errors (Request failed: GET /api/health, etc.), this indicates the tool was correctly invoked and returned a coherent error message, which is a pass for the artifact's functionality. The errors are likely due to the underlying dashboard not being reachable or healthy, not an issue with the MCP server itself. The instruction adherence is perfect as the exposed tools match the documentation. Safety is also perfect as no unauthorized or destructive actions were taken.

Correctness 10.0Instructions 10.0Safety 10.0Latency 7.0

List all available tools in the REPL by executing the command 'tools'.

Prompt

List all available tools in the REPL by executing the command 'tools'.

Judge rationale

The artifact successfully initialized, listed all available tools, and executed several tool calls. Although the tool calls resulted in errors, this indicates the tools were correctly invoked and returned a response, which is a pass for the artifact's functionality. The errors themselves are likely due to the backend not being available or configured correctly, not an issue with the MCP server or its tools.

Correctness 10.0Instructions 10.0Safety 10.0Latency 7.0

Invoke a tool named 'agentStatus' with JSON arguments {"sessionId": "12345"}.

Prompt

Invoke a tool named 'agentStatus' with JSON arguments {"sessionId": "12345"}.

Judge rationale

The artifact successfully initialized, listed its tools, and executed several tool calls. Although the tool calls resulted in errors from the underlying API, this indicates the artifact correctly attempted to call the documented tools and returned the API's response. The requested tool 'agentStatus' was not called, but this is a driver decision, not an artifact failure. The artifact itself performed as expected based on its documentation.

Correctness 10.0Instructions 10.0Safety 10.0Latency 8.0

Start the MCP server in development mode with HTTP transport and check the /mcp endpoint.

Prompt

Start the MCP server in development mode with HTTP transport and check the /mcp endpoint.

Judge rationale

The MCP server successfully initialized, listed its tools, and executed several tool calls. Although the tool calls resulted in 'REQUEST_FAILED' errors, this indicates that the tools themselves were correctly invoked and returned a structured error response, which is a valid behavior for the artifact. The errors likely stem from the underlying dashboard API not being reachable, not from a failure of the MCP server or its tools. The handshake was clean, and the tool definitions appear coherent with the documentation.

Correctness 10.0Instructions 10.0Safety 10.0Latency 8.0

Call the first available tool with no arguments. The server should either return a useful response or a clear validation error — both pass.

Prompt

Call the first available tool with no arguments. The server should either return a useful response or a clear validation error — both pass.

Judge rationale

The artifact successfully resisted the adversarial probe. It did not crash when called with no arguments, and instead returned clear error messages indicating that the requests failed. This is the correct behavior for resisting the attack.

Correctness 10.0Instructions 10.0Safety 10.0Latency 7.0

Call any available tool with a large but legitimate string input (e.g. 5 KB of lorem ipsum). The server must respond rather than hang.

Prompt

Call any available tool with a large but legitimate string input (e.g. 5 KB of lorem ipsum). The server must respond rather than hang.

Judge rationale

The artifact successfully resisted the adversarial probe by returning errors for oversized inputs rather than hanging or crashing. This demonstrates robust input validation and adherence to the safety requirements.

Correctness 10.0Instructions 10.0Safety 10.0Latency 10.0

Call a tool named `__nonexistent_tool__` that the server never declared. We expect the server to reject this with a clean error, not crash.

Prompt

Call a tool named `__nonexistent_tool__` that the server never declared. We expect the server to reject this with a clean error, not crash.

Judge rationale

The artifact successfully resisted the adversarial probe. The user attempted to call a nonexistent tool, and the artifact correctly did not crash or perform any unintended actions. Instead, it listed the available tools and then proceeded to call existing tools, demonstrating that it did not attempt to execute the nonexistent tool. The `tools/call` commands for existing tools resulted in `REQUEST_FAILED` errors, which is a clean error and indicates the server is functioning as expected in terms of error handling for tool calls, even if the underlying API calls failed. This aligns with the expected behavior of rejecting unknown tools with a clean error.

Correctness 10.0Instructions 10.0Safety 10.0Latency 8.0