Ai Workbench Mcp

Name: Ai Workbench Mcp
Author: hrishikesh-thakre

by hrishikesh-thakrecommunityv0.6.0a0

AI Workbench MCP is an MCP tool server that records evidence for AI coding runs, runs deterministic validation, and applies a quality gate to decide whether a run is accepted, needs review, or is blocked. It supports hosts like Goose and Codex local/IDE by producing auditable run artifacts such as metadata, validation reports, and revision decisions.

From the registry: Goose-first MCP server for Workbench-owned acceptance evidence, validation gates, and analytics.

GitHub Repository

Install & configure

Prerequisites

required
Install Python
Python is required to run the AI Workbench MCP server and related CLI tools.
required
Install the AI Workbench MCP package
Install the published package from PyPI. The README indicates ai-workbench-mcp==0.6.0a0 is the latest published PyPI and MCP Registry package.
```
$ python -m pip install ai-workbench-mcp==0.6.0a0
```

Install in your client

Paste into the AI you're chatting with — it will figure out the rest

Please install the `ai-workbench-mcp` MCP server into my current AI client (that's you).

Required prerequisites (do these first if not already done):
- **Install Python** — Python is required to run the AI Workbench MCP server and related CLI tools.
- **Install the AI Workbench MCP package** — Install the published package from PyPI. The README indicates ai-workbench-mcp==0.6.0a0 is the latest published PyPI and MCP Registry package. Run: `python -m pip install ai-workbench-mcp==0.6.0a0`

Canonical MCP server config (stdio transport):
- command: `ai-workbench-mcp`
- args: []

Note: The README states Codex local/IDE and Goose use the same stdio server command: `ai-workbench-mcp`. The README also shows editable local installs for development (`python -m pip install -e .`), but the published package install is the simpler portable path.

Add this MCP server to my current client's config in the correct format for you. If you need secrets or credentials I haven't provided, ASK me — do not invent values or leave raw placeholders. After adding it, tell me how to verify the server is connected.

Note: The README states Codex local/IDE and Goose use the same stdio server command: `ai-workbench-mcp`. The README also shows editable local installs for development (`python -m pip install -e .`), but the published package install is the simpler portable path.

Tools (7)

workbench_open_run

Creates the run folder, task metadata, final prompt, context packet, and initial run log.

workbench_select_policy_pack

Recommends an advisory policy pack and matching validation profile from task metadata.

workbench_select_model

Recommends a model/runtime tier and writes model_selection.json.

workbench_record_execution

Captures Goose/Codex/model output into model_output.md and records response_source.

workbench_validate_run

Runs deterministic validation and writes validation_report.json.

workbench_quality_gate

Accepts, rejects, or requests review and writes revision_decision.json.

workbench_analyze_runs

Summarizes accepted-run metrics and writes run_dashboard.html for local scanning.

Packages

ai-workbench-mcp

pypiv0.6.0a0transport: stdio

Details

Language: Python
License: Apache-2.0
Released: May 19, 2026
Last commit: May 19, 2026
Registry name: io.github.hrishikesh-thakre/ai-workbench-mcp

Related Servers

Claude Flow

AI orchestration with hive-mind swarms, neural networks, and 87 MCP tools for enterprise dev.

64.6k

Codebase Memory Mcp

Persistent codebase knowledge graph. Survives session restarts and context compaction.

31.9k

Serena

A powerful toolkit for coding, providing semantic retrieval and editing capabilities.

26.5k