// SCAN REPORT

Workbench

15 tools. 11 can modify or destroy data without limits.

11 write tools that can modify data. Rate limits recommended.

Last updated: 8 Jul 2026

11 can modify or destroy data

4 read-only

15 tools total

Community server · catalogue entry checked 08/07/2026

How to control Workbench ↓

TOOL MAP

What Workbench exposes to your agents

Read (4) Write / Execute (11) Destructive / Financial (0)

Running more than one server? Check your whole stack →

MOST DANGEROUS TOOLS

High Risk

The most dangerous Workbench tools

Execute · High wb_sync This tool performs a full reconciliation operation ('全量对账') that includes invalidation detection ('失效检测'), dispatching reviews ('review 派发'), and creating tombstones ('tombstone'). Write · Medium wb_claim This tool claims/assigns a task to an agent, which modifies task state (ownership/status) and creates snapshots of upstream dependencies. Write · Medium wb_feedback The tool creates or modifies feedback records that influence artifact approval decisions in a multi-stage pipeline. Write · Medium wb_input This tool creates or registers new task dependency declarations and modifies monitoring state for artifacts.

11 of Workbench's 15 tools can modify, destroy, or commit something on every call — and an agent calls them with no built-in limits.

POLICY

How to control Workbench

PolicyLayer is an MCP gateway — it sits between your AI agents and Workbench, and nothing reaches the server without passing your rules. These are the rules we recommend:

Rate limit write operations

{
  "wb_claim": {
    "limits": [
      {
        "counter": "wb_claim_per_hour",
        "window": "hour",
        "max": 30,
        "scope": "grant"
      }
    ]
  }
}

Prevents bulk unintended modifications from agents caught in loops.

Cap read operations

{
  "wb_artifacts": {
    "limits": [
      {
        "counter": "wb_artifacts_per_minute",
        "window": "minute",
        "max": 60,
        "scope": "grant"
      }
    ]
  }
}

Controls API costs and prevents retry loops from exhausting upstream rate limits.

Create a free account and register Workbench — nothing to install.
Add these rules — paste them, or build them visually. Tune the limits to your setup.
Point your MCP client (Claude, Cursor, anything) at your gateway URL.

ENFORCE POLICY ON WORKBENCH →

Instant setup, no code required.

FULL CATALOGUE

All 15 Workbench tools

EXECUTE 1 tools

Execute wb_sync 全量对账:失效检测+review 派发+tombstone

WRITE 10 tools

Write wb_claim 领取任务(gate 自动校验,上游依赖自动快照) Write wb_dispute 对 approved 内容有实质异议时留痕并停止(等用户裁决,禁止擅自偏离) Write wb_feedback 对原型/代码产物记录 👍(+1)/👎(-1);👎 必附 comment;原型 👍=放行 Write wb_input 补充申报任务依赖(gate 之外读过的登记产物,进入 stale 监控) Write wb_output 登记产出文件(必须先写文件;自动推断 kind 并关联已领取任务) Write wb_plan 按已登记 page-prd 派发模块整组任务(幂等,删页自动 cancel) Write wb_qa 记录 QA 验收结果;fail 必附 reason(自动派 rework,完成后自动复验) Write wb_record 任务备注留痕 Write wb_submit 契约文档送审(进入用户待审队列) Write wb_update_status 更新任务状态;completed 会跑 stale 拦截/产出义务/协议 lint/机器检查

READ 4 tools

Read wb_artifacts 查看产物信任状态(approved=真相直接用;invalidated/复审中=禁用) Read wb_audit 模块对账报告(清算状态/契约信任状态/送审建议) Read wb_list_tasks 查看任务列表(gate 报错可行动;领取用 wb_claim) Read wb_retro retrospective 证据包:反馈半衰期加权提炼(skill 候选/Red Flags/观察)+ 审批吞吐报表。

FAQ

Questions about Workbench

How do I prevent bulk modifications through Workbench? +

The Workbench server has 10 write tools including wb_claim, wb_dispute, wb_feedback. Set a rate limit in your policy -- for example, 10 calls per hour prevents an agent from making more than 10 modifications per hour. PolicyLayer enforces this at the gateway, before calls reach Workbench.

How many tools does the Workbench MCP server expose? +

15 tools across 2 categories: Read, Write. 4 are read-only. 11 can modify, create, or delete data.

How do I enforce a policy on Workbench? +

Register the Workbench MCP server in PolicyLayer, apply the suggested rules above (adjust the limits to your use case), and point your AI client at the PolicyLayer proxy URL instead of the server directly. Your agents keep the same tools; PolicyLayer evaluates every call against policy before it executes. Nothing to install, live in minutes.

Enforce policy on every Workbench tool call.

Deterministic rules across all 15 Workbench tools. Per-identity grants. Full audit log. Live in minutes. Nothing to install.

ENFORCE POLICY ON WORKBENCH →

Instant setup, no code required.

15 Workbench tools catalogued and risk-classified — across an index of 46,500+ MCP servers.

// WHERE THIS COMES FROM

These policies come from Workbench's registry record.

The record behind this page: verified identity, auth posture, risk grade, every tool classified, recommended policy — re-checked continuously.

Teams ship this data inside their own products. See what a licence covers →