en es ja ru zh

errorbar

Name: errorbar
Author: omnia-v

by omnia-v

Overview Schema Related Servers Score Discussions

TypeScript

Remote

@omnia-voice/mcp

将 errorbar 平台封装为 MCP 工具——所有公开的 API 操作（evals、judges/criteria、gates、logs、datasets、aliases、label sets、audit/proving、dedicated endpoints、training）都可以从 Claude Code、Claude Desktop、Cursor 或任何 MCP 客户端调用。

每个 API 操作对应一个工具，共 76 个工具，全部由平台的路由契约生成。输入按 REST API 接受的字段名和大小写进行校验；响应与原 API 返回完全一致（JSON 美化缩进，导出为文本），包括 API 自身的错误状态和错误消息。

安装

npx -y @omnia-voice/mcp        # or: npm i -g @omnia-voice/mcp && omnia-mcp

需要 OMNIA_API_KEY（工作区 API 密钥，形如 sk_…，在 设置 → API 密钥 中获取）。只给该密钥授予代理应拥有的作用域——监控代理给只读的 read，运行 evals 的代理给它 read + evals:write——平台会强制校验这些作用域，且每个工具的描述都会注明它需要哪个作用域。

Claude Code

claude mcp add errorbar -e OMNIA_API_KEY=sk_… -- npx -y @omnia-voice/mcp

Claude Desktop / Cursor / 任意 `mcpServers` 配置

{
  "mcpServers": {
    "errorbar": {
      "command": "npx",
      "args": ["-y", "@omnia-voice/mcp"],
      "env": { "OMNIA_API_KEY": "sk_…" }
    }
  }
}

Related MCP server: aman-mcp

标志

标志	效果
`--read-only`（或 `OMNIA_MCP_READ_ONLY=1`）	只暴露 `GET` 工具
`--no-spend`（或 `OMNIA_MCP_NO_SPEND=1`）	隐藏会启动计费工作的工具（eval 运行、训练、专用容量、judge 辅助）
`--only list_evals,get_eval`	只暴露这几个工具
`--base-url URL`（或 `OMNIA_BASE_URL`）	API 根地址，默认为 `https://platform.omnia-voice.com/api`

需要消耗钱包余额的工具会在描述中标注 SPENDS MONEY，且不会被标记为 readOnlyHint；MCP 客户端若在非只读调用前请求确认，这些工具也会触发确认。

工具

工具	操作	作用域
`list_alerts`	`GET /v1/alerts`	read
`list_aliases`	`GET /v1/aliases`	read
`upsert_alias`	`PUT /v1/aliases`	aliases:write
`delete_alias`	`DELETE /v1/aliases/{id}`	aliases:write
`export_audit_log`	`GET /v1/audit/export`	read
`list_audit_tombstones`	`GET /v1/audit/tombstones`	read
`create_audit_tombstone`	`POST /v1/audit/tombstones`	platform:write
`get_audit_verification`	`GET /v1/audit/verify`	read
`list_batches`	`GET /v1/batches`	read
`create_batch`	`POST /v1/batches`	platform:write · 计费
`get_batch`	`GET /v1/batches/{id}`	read
`cancel_batch`	`POST /v1/batches/{id}/cancel`	platform:write
`list_criteria`	`GET /v1/criteria`	read
`create_criterion`	`POST /v1/criteria`	evals:write
`suggest_criteria`	`POST /v1/criteria/suggest`	evals:write · 计费
`list_criterion_templates`	`GET /v1/criteria/templates`	read
`get_criterion`	`GET /v1/criteria/{id}`	read
`update_criterion`	`PATCH /v1/criteria/{id}`	evals:write
`delete_criterion`	`DELETE /v1/criteria/{id}`	evals:write
`run_criterion_alignment`	`POST /v1/criteria/{id}/align`	evals:write · 计费
`get_criterion_alignment`	`GET /v1/criteria/{id}/alignment`	read
`auto_improve_criterion`	`POST /v1/criteria/{id}/auto_improve`	evals:write · 计费
`get_criterion_certificate`	`GET /v1/criteria/{id}/certificate`	read
`scan_criterion_suspects`	`POST /v1/criteria/{id}/scan`	evals:write · 计费
`decontaminate_texts`	`POST /v1/datasets/decontaminate`	evals:write
`create_dataset_from_logs`	`POST /v1/datasets/from_logs`	evals:write
`list_dedicated_endpoints`	`GET /v1/dedicated`	read
`create_dedicated_endpoint`	`POST /v1/dedicated`	platform:write · 计费
`list_dedicated_templates`	`GET /v1/dedicated/templates`	read
`get_dedicated_endpoint`	`GET /v1/dedicated/{id}`	read
`update_dedicated_endpoint`	`PATCH /v1/dedicated/{id}`	platform:write · 计费
`delete_dedicated_endpoint`	`DELETE /v1/dedicated/{id}`	platform:write
`list_refusals`	`GET /v1/enforcement/refusals`	read
`list_env_tools`	`GET /v1/env/tools`	read
`register_env_tool`	`POST /v1/env/tools`	platform:write
`delete_env_tool`	`DELETE /v1/env/tools/{id}`	platform:write
`list_evals`	`GET /v1/evals`	read
`create_eval`	`POST /v1/evals`	evals:write · 计费
`compare_evals`	`GET /v1/evals/compare`	read
`get_failure_clusters`	`GET /v1/evals/failure_clusters`	read
`get_eval`	`GET /v1/evals/{id}`	read
`delete_eval`	`DELETE /v1/evals/{id}`	evals:write
`cancel_eval`	`POST /v1/evals/{id}/cancel`	evals:write
`get_eval_evidence`	`GET /v1/evals/{id}/evidence`	read
`get_eval_gate`	`GET /v1/evals/{id}/gate`	read
`get_eval_samples`	`GET /v1/evals/{id}/samples`	read
`list_training_files`	`GET /v1/fine_tuning/files`	read
`upload_training_file`	`POST /v1/fine_tuning/files`	platform:write
`list_fine_tuning_jobs`	`GET /v1/fine_tuning/jobs`	read
`create_fine_tuning_job`	`POST /v1/fine_tuning/jobs`	platform:write · 计费
`get_fine_tuning_job`	`GET /v1/fine_tuning/jobs/{id}`	read
`cancel_fine_tuning_job`	`DELETE /v1/fine_tuning/jobs/{id}`	platform:write
`get_fine_tuning_bakeoff`	`GET /v1/fine_tuning/jobs/{id}/bakeoff`	read
`start_fine_tuning_bakeoff`	`POST /v1/fine_tuning/jobs/{id}/bakeoff`	platform:write · 计费
`list_grpo_runs`	`GET /v1/grpo/runs`	read
`start_grpo_run`	`POST /v1/grpo/runs`	platform:write · 计费
`get_grpo_run`	`GET /v1/grpo/runs/{id}`	read
`stop_grpo_run`	`POST /v1/grpo/runs/{id}/stop`	platform:write
`get_grpo_run_weights`	`GET /v1/grpo/runs/{id}/weights`	read
`list_label_sets`	`GET /v1/label_sets`	read
`create_label_set`	`POST /v1/label_sets`	evals:write
`attach_label_set`	`POST /v1/label_sets/{id}/attach`	evals:write
`freeze_label_set`	`POST /v1/label_sets/{id}/freeze`	evals:write
`list_labels`	`GET /v1/labels`	read
`create_label`	`POST /v1/labels`	evals:write
`list_logs`	`GET /v1/logs`	read
`export_logs`	`GET /v1/logs/export`	read
`list_model_versions`	`GET /v1/model_versions`	read
`get_model_version`	`GET /v1/model_versions/{id}`	read
`adopt_model_version`	`POST /v1/model_versions/{id}/adopt`	platform:write
`list_raft_rounds`	`GET /v1/raft/rounds`	read
`get_judge_settings`	`GET /v1/settings/judge`	read
`set_judge_settings`	`PUT /v1/settings/judge`	platform:write
`get_setup_status`	`GET /v1/setup/status`	read
`get_trace`	`GET /v1/traces/{traceId}`	read
`verify_document`	`POST /v1/verify`	read

每个工具的描述都附带了完整契约：所有的 query 和 body 字段、响应包含什么内容，以及常见的坑（feature flags、400 条件、cursor 分页语义）。

以编程方式使用

import { createServer, OmniaClient } from "@omnia-voice/mcp";
import { StdioServerTransport } from "@modelcontextprotocol/sdk/server/stdio.js";

const server = createServer({ client: new OmniaClient({ apiKey: process.env.OMNIA_API_KEY! }), readOnly: true });
await server.connect(new StdioServerTransport());

开发

npm ci && npm run typecheck && npm test && npm run build

src/manifest.ts 由平台的 app/api/v1 路由生成；test/manifest.test.ts 固定了其必须覆盖的精确路由清单，因此只要出现新的平台路由而清单未重新生成，这套测试就会失败。

发布

推送一个 vX.Y.Z 标签——发布工作流会构建、测试，并携带 provenance 发布到 npm。

许可证：Apache-2.0

This server cannot be installed

license - permissive license

Not graded

quality - not tested

maintenance

How are these scores calculated?

Maintenance

–Maintainers

–Response time

–Release cycle

–Releases (12mo)

Commit activity

Resources

GitHub Repository

Need Help?

Related Servers

Unclaimed servers have limited discoverability.

Looking for Admin?

If you are the server author, to access and configure the admin panel.

Related MCP Servers

opendata-mcp
Open Data Data Platforms
Skeego
F
license
C
quality
D
maintenance
Enables interaction with the OpenData Platform API by dynamically exposing all endpoints as MCP tools with typed input schemas and HTTP handlers.
Updated 3 months ago (2026-05-20 03:08 UTC)
99
aman-mcp
Agent Orchestration Developer Tools Autonomous Agents
amanasmuei
A
license
Not graded
quality
B
maintenance
Exposes identity, tools, workflows, guardrails, and evaluation as MCP tools — so any AI agent can read and write your ecosystem programmatically.
Updated 3 months ago (2026-06-10 07:28 UTC)
32
MIT
JustOneAPI MCP Serverofficial
Search Social Media
justoneapi
A
license
Not graded
quality
A
maintenance
Exposes JustOneAPI endpoints as MCP tools, returning raw upstream JSON without field parsing for maximum data fidelity.
Updated 2 days ago (2026-08-25 10:43 UTC)
16
28
MIT
Patronus MCP Serverofficial
AI & Machine Learning Testing & QA Tools
patronus-ai
A
license
Not graded
quality
D
maintenance
Enables running LLM evaluations, experiments, and custom evaluators through a standardized MCP interface.
Updated a year ago (2025-03-26 12:33 UTC)
16
Apache 2.0

View all related MCP servers

Related MCP Connectors

DATPAQ
34 production API tools over one hosted MCP endpoint.
coralquery-mcp
A paid remote MCP for AI SDK data query MCP, built to return verdicts, receipts, usage logs, and aud
evalscopebench-mcp
A paid remote MCP for AI SDK benchmark dashboard, built to return verdicts, receipts, usage logs, an

View all MCP Connectors

Latest Blog Posts

Who's Calling? MCP Hosts Are an Identity Blind Spot (And the Spec Knows It)
By Om-Shree-0709 on July 25, 2026.
mcp
Agent Identity
OAuth 2.1
Your AI Chatbot Just Exposed Your CEO's Salary to an Intern
By Om-Shree-0709 on July 2, 2026.
Agent Identity
MCP Security
OAuth Delegation
Why MCP Servers Need Execution Sandboxing (And Why Your Current Stack Isn't Enough)
By Om-Shree-0709 on June 30, 2026.
Agentic Ai
Prompt Injection
WebAssembly

MCP directory API

We provide all the information about MCP servers via our MCP API.

curl -X GET 'https://glama.ai/api/mcp/v1/servers/omnia-v/omnia-mcp'

If you have feedback or need assistance with the MCP directory API, please join our Discord server

@omnia-voice/mcp

安装

Claude Code

Claude Desktop / Cursor / 任意 mcpServers 配置

标志

工具

以编程方式使用

开发

发布

Maintenance

Resources

Looking for Admin?

Related MCP Servers

opendata-mcp

aman-mcp

JustOneAPI MCP Serverofficial

Patronus MCP Serverofficial

Related MCP Connectors

Latest Blog Posts

MCP directory API

Claude Desktop / Cursor / 任意 `mcpServers` 配置