AI Experiment Logger MCP Server

Community

MCP server for logging and analyzing AI experiments

How to install ↓ Compare GitHub source repository ↗

Category: Dev Tools #327 of 438
Stars: ★ 2.9k Very popular
Transport: stdio (local process)
Runtime: Node.js
Credentials: No credential needed
License: MIT
Last commit: 1 day ago
Tools: 7

43FMRS · D

ai-experiment-logger is a Node.js-based stdio MCP server offering 7 tools for logging, querying, updating, deleting, summarizing, and exporting AI experiment data. It ships from a large community plugin marketplace repository, is not officially maintained, and suits individuals or small teams doing lightweight experiment tracking, though data sensitivity and supply-chain provenance warrant user diligence.

Strongest · Maintenance 13/20 Weakest · Documentation 5/20

Reliability

7/20

Security and permissions

11/20

Maintenance

13/20

Documentation

5/20

Setup experience

7/20

Why each score

Reliability 7/20: The only evidence is the server.json's declared command/args and a list of 7 tool names (log_experiment, list_experiments, get_experiment, update_experiment, delete_experiment, get_statistics, export_experiments). No test files, CI workflow, or execution logs for this specific server directory are provided, so whether declared tools match actual behavior cannot be verified. The README's plugin-table description ('web dashboard and MCP tools') differs slightly in wording from the manifest description ('comprehensive logging and statistics'), adding uncertainty. Per the static calibration rule, reliability cannot exceed 12 without verifiable execution evidence (real CI plus tests covering key paths); repo-level mentions of a marketplace validator apply to the whole catalog, not server-specific tests. Scored 7: a plausible happy path can be inferred from the tool names, but error handling, dependency management, and edge cases are entirely opaque.
Security and permissions 11/20: The manifest discloses no permission scope, credential handling, storage location, or external network behavior. Notably, the tool list includes delete_experiment (a destructive operation) with no evidence of a confirmation step before irreversible deletion, and export_experiments has no disclosed destination/scope for exported data. There is no indication of data exfiltration, payment, or remote-command execution, so this does not rise to red-line territory (0-4), but it is a clear undisclosed-risk gap. Scored 11: main risk areas (delete/export) are visible in the tool list, but scoping, confirmation, and data-boundary disclosure are all incomplete.
Maintenance 13/20: The repository as a whole shows credible maintenance signals: MIT license, v4.33.0 release, actively updated npm download stats (daily GitHub Actions cron), 30 community contributors, and a public validator/grading rubric. However, these signals describe governance of the entire marketplace repo, not maintenance activity specific to the ai-experiment-logger server (no server-specific issue history, changelog, or dependency-update record is provided). Scored 13: clear repo-level governance and licensing, but no server-specific maintenance evidence, which caps the score short of full marks.
Documentation 5/20: Documentation for this server is limited to a single description line in the README's category table; there is no install procedure, authentication requirement, per-tool parameter documentation, usage limits, cost information, or troubleshooting guidance specific to ai-experiment-logger. The repo's Quick Start section documents generic marketplace installation (ccpi CLI, /plugin marketplace add) rather than this server's specific mcpServers connection setup. Documentation is substantially deficient, scored 5.
Setup experience 7/20: The manifest path is a placeholder — '/absolute/path/to/plugins/mcp/ai-experiment-logger/dist/index.js' — not a working real path, meaning the user must locate and substitute it manually; this is not a ready-to-use config example. No package.json, build script, or instructions for producing dist/index.js are shown, and there is no server-specific client configuration example. Per the static calibration rule, setup cannot exceed 15 without verifiable execution evidence; given the placeholder path and lack of dedicated setup documentation, scored 7.

Static review · not runListed 2026-08-06

Read the FMRS scoring method →

Fit and risk

What it can accessConnects to a database

Best for

Developers who want lightweight tracking of AI experiments inside a Claude Code workflow
Users who prefer MCP tools over manual spreadsheets for managing experiment metadata

Not for

Teams needing an enterprise-grade experiment tracking platform with versioning, collaboration, or cloud sync
Production deployments requiring official Anthropic support or independent security audits

Required permissions

Local filesystem read/write access to persist experiment log data
Ability to run a Node.js process communicating with the MCP client over stdio

Risks and side effects

This server comes from a community marketplace repository with hundreds of plugins; it is not maintained by Anthropic and has no stated independent security audit
Experiment records may contain sensitive data such as prompts or outputs, so local storage access should be controlled
Installing via a third-party CLI (ccpi) or marketplace script carries some supply-chain risk; review the source before installing

Setup

Before you start

Runtime：Node.js

Build the server from plugins/mcp/ai-experiment-logger in the repository to produce dist/index.js (requires Node.js), then add a stdio MCP server entry in your client config with command set to node and args pointing to the absolute path of that dist/index.js file. Alternatively, install via the repository's ccpi CLI or Claude Code's /plugin marketplace add command to pull in the plugin marketplace before enabling this server.

claude_desktop_config.json

{"mcpServers":{"ai-experiment-logger":{"command":"node","args":["/absolute/path/to/plugins/mcp/ai-experiment-logger/dist/index.js"]}}}

Shown for Claude Desktop. Other clients may use a different file or key (VS Code uses "servers") — the configurator below converts it.

.vscode/mcp.json

{
  "servers": {
    "ai-experiment-logger": {
      "command": "node",
      "args": [
        "/absolute/path/to/plugins/mcp/ai-experiment-logger/dist/index.js"
      ]
    }
  }
}

Goes in your project's .vscode/mcp.json (VS Code uses a "servers" key).

Terminal

claude mcp add ai-experiment-logger -- node /absolute/path/to/plugins/mcp/ai-experiment-logger/dist/index.js

Run it in a terminal; replace any <…> placeholders with your own values first.

Generate and validate client config

Check that it works

After installing and adding ai-experiment-logger to your client config, the tool list should show log_experiment, list_experiments, get_statistics and the other recorded tools; calling list_experiments and getting a normal response confirms the connection works.

Troubleshooting

Confirm the build step has been run and dist/index.js actually exists
Verify the path in the MCP config is an absolute path matching the actual build output location
Confirm a supported Node.js version is installed
Check MCP client logs to verify the stdio connection is established correctly
Ensure the process running the server has read/write permission to the experiment data storage location

Things to try

Once connected, you can ask your AI assistant things like:

Log an experiment named baseline-run with its parameters and results.
List all experiments I have recorded so far.
Show me statistics across all my logged experiments.
Delete the experiment named baseline-run and then export the remaining experiments.

Tools 7

log_experiment writes

Record a new AI experiment along with its parameters and results

list_experiments read-only

List all logged AI experiments

get_experiment read-only

Retrieve the details of a specific experiment

update_experiment writes

Update an existing experiment record

delete_experiment destructive

Delete a specific experiment record

get_statistics read-only

Compute summary statistics from logged experiments

export_experiments read-only

Export experiment data

Use cases

Log parameters and results from repeated AI/model experiments for later comparison

Query, update, or delete historical experiment records

Generate aggregate statistics across logged experiments

Export experiment data for reporting or further analysis

Supported clients

Claude Desktop

Claude Code

Listed from the project's documentation, not tested by this site.

Overview

ai-experiment-logger is an MCP server plugin from the claude-code-plugins-plus-skills repository, running locally over stdio via Node.js. It provides tools to log, query, update, delete, summarize, and export AI experiment records. The repository is an open-source Claude Code plugin and skill marketplace (tonsofskills.com); this server is one of 8 MCP servers bundled in it, released under the MIT license and not maintained by Anthropic.

Similar servers

Context7 80 · B

Upstash's official server providing up-to-date third-party library docs for AI coding assistants

★ 64.1k · Tools 2 Compare with this →

NPM Sentinel MCP Server 79 · B

AI-powered NPM package analysis MCP server

★ 18 · Tools 19 Compare with this →

MartinLoop MCP Server 79 · B

An independent control layer that gives AI coding agents budgets, brakes, and receipts

★ 394 Compare with this →

GitHub MCP Server 76 · B

GitHub's official MCP server for managing repos, issues, PRs, and workflows via natural language

★ 34.2k · Tools 15 Compare with this →

Compare with other MCP servers

More servers from this repository

Source revision eb6b1890fddf Data synced 2026-11-04 Read the FMRS scoring method