Bench Agent Discovery

io.github.VirajMishra1由 GitHub 帳號 VirajMishra1 發布v0.1.0更新於 Oct 24, 2026

Discover public AI agents, reusable recipes, and trusted benchmark evidence by task.

已驗證Streamable HTTP可網頁執行知識與記憶AI 與機器學習

概覽

AI 產生的概覽

一個公開、唯讀的 MCP 伺服器，用來依任務探索 AI 代理、其可重複使用的配方，以及可信的基準評測證據。

可以做什麼
依任務、能力、類別、框架、模型、已驗證證據或重用設定搜尋已列出的公開代理，擁有者遙測與基準證據會分開回傳。取得單一公開代理的配方、公開能力清單、粗略呼叫狀態、擁有者遙測和已驗證的基準提交。列出公開、具版本的基準合約，以及僅由受信任執行器驗證的提交。
適用情境
適合在希望助理為某項任務尋找現成的公開代理、查看某個代理的設定方式，或在採用或比較代理前確認哪些基準結果經過驗證時使用。
執行需求
遠端 streamable HTTP 端點；未宣告需要驗證、環境變數或標頭。需要能連線至該託管服務的網路。
安裝前請注意
僅供唯讀探索，不會執行或安裝代理。結果只涵蓋公開代理，擁有者自行回報的遙測與已驗證的基準證據分開標示，對自報數據應謹慎看待。

安裝

在 SourceWeft 中

  1. 開啟 儀表板中的 Bench Agent Discovery，將其新增到工作區。
  2. 為需要使用其工具的對話啟用該服務。

Web executable，透過 Streamable HTTP。 遠端服務在工作區中設定後即可從網頁執行環境執行。

交給代理程式安裝

把這段提示詞貼上到 Claude Code、Codex、Cursor 等代理程式，或 SourceWeft 對話中。代理程式會讀取該伺服器的安裝指南，先向你展示發布者、端點和登入方式，經你同意後再把它新增到你的客戶端。金鑰不會寫進任何命令。

閱讀 https://sourceweft.com/mcp/io.github.VirajMishra1%2Fbench/install.md，並按其中說明把這個 MCP 伺服器接入我的客戶端。修改任何設定之前，先告訴我它的發布者、端點和登入方式，等我確認。不要把 API 金鑰寫進命令或我會提交的檔案；問我金鑰應從哪裡取得。

檢視代理程式所依據的安裝指南

其他 MCP 客戶端

把它新增到你客戶端的 mcpServers 設定中。

{
  "mcpServers": {
    "bench-agent-discovery": {
      "type": "http",
      "url": "https://bench.virajmishratakehome.workers.dev/mcp"
    }
  }
}

README

bench

See what the best agents do differently.

One line of code. Live dashboard, public profile, README badge.

[Live] [npm] [PyPI] [License] [Built on]

[Bench dashboard — live event stream, task history, eval scores, README badge]

What is this?

You built an AI agent. You ran it a few times. But you have no idea if it's actually working well — which tasks fail silently, what it costs per run, or how it compares to anything else.

Bench fixes that. Wrap your agent with one function call. You get:

  • A public profile page showing runs, success rate, cost, and latency
  • An auto-score on every task (0–1, LLM-as-judge)
  • AI-generated summaries of your failure patterns
  • A README badge that stays live and updates as your agent runs
  • A public leaderboard so anyone can discover your agent

It's like GitHub for agents — observable, shareable, and public by default.

Want to see it before signing up? Try the sandbox at /try — no signup needed.

Setup (3 minutes)

Sign in at bench.virajmishratakehome.workers.dev with GitHub. The dashboard gives you a copyable setup bundle — install command, API key, and first task template. It listens for your first event and links straight to your profile when it arrives.

Or do it manually:

bash
npm install @virajmishra1/bench-sdkexport BENCH_KEY="bk_..."
typescript
import { observe } from "@virajmishra1/bench-sdk";
const agent = observe({ apiKey: process.env.BENCH_KEY, agent: "my-agent" });
await agent.task("search", { query }, async (t) => {  const result = await doSearch(query);  t.log("found", result.length);  t.cost(0.004);  return result;});

That's the whole SDK. Everything else is optional.

Python:

bash
pip install bench-observeexport BENCH_KEY="bk_..."
python
import bench
agent = bench.observe(api_key=os.environ["BENCH_KEY"], agent="my-agent")
async with agent.task_ctx("search", {"query": query}) as task:    result = await do_search(query)    task.log("found", len(result))    task.set_output(result)

Already on OpenTelemetry? Point your exporter at Bench instead:

bash
export OTEL_EXPORTER_OTLP_ENDPOINT=https://bench.virajmishratakehome.workers.devexport OTEL_EXPORTER_OTLP_PROTOCOL=http/jsonexport OTEL_EXPORTER_OTLP_HEADERS="X-Bench-Key=bka_...,X-Bench-Agent=my-agent"

Bench understands standard gen_ai.* spans — invoke_agent, execute_tool, chat, retrieval, and more.

Prefer a CLI? The stack-detecting CLI auto-instruments OpenAI, Anthropic, Vercel AI SDK, Mastra, and LangChain:

bash
npx @virajmishra1/bench-cli init --installnpx @virajmishra1/bench-cli login

What you get

FeatureDescription
Live dashboardReal-time event stream while your agent runs. WebSocket, zero polling.
Public profile/u/you/your-agent — shareable, OG-image ready, server-rendered
README badgeLive SVG badge. Updates automatically. GitHub camo-friendly.
LLM evalEvery task auto-scored 0–1 by a Llama 3.3 70B judge. Score logic is open.
Failure insightsk-means clustering + LLM description of what keeps going wrong
LeaderboardBrowse public agents by runs, success rate, eval score, or cost
Compare/vs/@a/agent1/@b/agent2 — side-by-side quality, cost, latency
BenchmarksVersioned benchmark suites with repeated runs and evidence trails. Separate from self-reported telemetry.
MCP discoveryPublic read-only MCP server — search_agents, get_agent, list_benchmarks
Embed widget<iframe>-ready mini-dashboard, 3 sizes, dark/light
Privacy controlsHide inputs/outputs, make agents private, per-key access
Permissioned reusePublish capabilities with deny-by-default policies and quotas

Framework adapters

Drop-in wrappers that auto-instrument your existing LLM calls:

typescript
// Anthropic — wraps every messages.create() callimport { wrapAnthropic } from "@virajmishra1/bench-anthropic";const client = wrapAnthropic(new Anthropic(), bench);
// OpenAI — wraps chat completions, responses, and embeddingsimport { wrapOpenAI } from "@virajmishra1/bench-openai";const client = wrapOpenAI(new OpenAI(), bench);
// Vercel AI SDK — wraps generateText / streamText / generateObjectimport { track } from "@virajmishra1/bench-vercel-ai";const result = await track(bench, "summarize", () =>  generateText({ model: anthropic("claude-sonnet-4-6"), prompt: "..." }));
// Mastraimport { wrapMastra } from "@virajmishra1/bench-mastra";

Let your AI find agents

Bench exposes a public MCP server at /mcp. Connect it to Claude Code:

bash
claude mcp add --transport http bench https://bench.virajmishratakehome.workers.dev/mcp

Or Codex:

bash
codex mcp add bench --url https://bench.virajmishratakehome.workers.dev/mcp

Tools available: search_agents, get_agent, list_benchmarks. Search returns only public agents. Owner telemetry and benchmark evidence are labeled separately.

See MCP.md for full tool schemas and the privacy model.

Architecture

Bench runs entirely on Cloudflare. Each product is doing a specific job:

SDK (npm: @virajmishra1/bench-sdk)        |  batched events, X-Bench-Key        vPOST /ingest                              <- Workers (Hono)        |        +-> D1 --- users, agents, tasks, events        |        +-> AgentDO --- one Durable Object per agent        |           +- ring buffer (last 1k events, SQLite in DO storage)        |           +- latency histogram (p50, p95)        |           +- hibernating WebSocket -> live dashboards        |        +-> EvalWorkflow --- runs per task.end        |           +- Workers AI (Llama 3.3 70B) -> score 0-1 + reasoning        |              -> writes back to D1.tasks        |              -> updates agents.avg_eval_score        |        +-> ClusterWorkflow --- on-demand + hourly cron                    +- k-means on task embeddings -> cluster labels                       -> Workers AI LLM describes each cluster                       -> stored in agents.failure_clusters
Public surfaces:  /u/:login/:slug          -> profile page (server-rendered, OG image)  /badge/:login/:slug.svg  -> README badge (KV-cached 60s)  /embed/:login/:slug      -> iframe widget (3 sizes, dark/light)  /leaderboard             -> discovery (5 sort modes)  /vs/:a/:b                -> compare two agents  /try                     -> sandbox (no signup)  /benchmarks              -> verified benchmark registry  /mcp                     -> read-only MCP server  /api/agents/:l/:s/insights -> failure pattern analysis (JSON)

The key design decision is the actor model: every agent gets its own Durable Object. That DO holds the last 1,000 events in SQLite, a latency histogram, and a hibernating WebSocket connection — zero idle cost, no polling.

Cloudflare products used

ProductRole
WorkersAPI, profile rendering, badge generation
Durable ObjectsOne per agent — ring buffer, latency histogram, hibernating WebSocket
D1Users, agents, tasks, events
KVToken lookup cache, badge SVG cache, OG image cache
Workers AILlama 3.3 70B — LLM judge + failure pattern descriptions
WorkflowsDurable retry for EvalWorkflow and ClusterWorkflow
Browser RenderingOG share images (SVG → PNG)
AssetsStatic frontend (landing, dashboard, JS, CSS)

SDK reference

typescript
const agent = observe({  apiKey: string;           // bk_xxx — from your dashboard  agent: string;            // slug, e.g. "my-agent"  displayName?: string;  endpoint?: string;        // default: bench.virajmishratakehome.workers.dev  flushIntervalMs?: number; // default: 2000  maxBatchSize?: number;    // default: 50});
// Wrap a task — records start/end/duration/status/eval automaticallyawait agent.task("name", input, async (task) => {  task.log("label", value);    // attach a log event  task.cost(0.003);            // report LLM spend (owner-reported)  return result;               // returned value becomes the task output});
// Fire a custom eventagent.event("custom", { key: "value" });
// Flush immediately (auto-runs on batch full or interval)await agent.flush();

task.cost() calls are labeled "owner-reported" in the UI. Framework adapters attach provider and token evidence, labeled separately.

Errors are swallowed silently — observability should never crash your agent.

Self-host

bash
git clone https://github.com/VirajMishra1/benchcd bench && npm install
cd packages/workernpx wrangler login
# Create infrastructurenpx wrangler d1 create bench-dbnpx wrangler kv namespace create CACHEnpx wrangler kv namespace create SESSIONS
# Paste the returned IDs into wrangler.jsonc, then:npx wrangler secret put SESSION_SECRET             # any random 32+ char stringnpx wrangler secret put GITHUB_OAUTH_CLIENT_SECRET # from github.com/settings/developers
# Apply schema and deploynpm run db:remotenpm run deploy

File layout

bench/+-- packages/|   +-- sdk/                   <- @virajmishra1/bench-sdk|   +-- adapters/|   |   +-- anthropic/         <- @virajmishra1/bench-anthropic|   |   +-- openai/            <- @virajmishra1/bench-openai|   |   +-- vercel-ai/         <- @virajmishra1/bench-vercel-ai|   |   +-- mastra/            <- @virajmishra1/bench-mastra|   |   +-- langchain/         <- bench-langchain|   +-- worker/                <- Cloudflare Worker (all backend + frontend)|       +-- src/|       |   +-- index.ts       <- Hono routes|       |   +-- ingest.ts      <- POST /ingest|       |   +-- profile.ts     <- public profile page|       |   +-- badge.ts       <- SVG README badge|       |   +-- embed.ts       <- iframe widget|       |   +-- leaderboard.ts <- discovery page|       |   +-- compare.ts     <- /vs/:a/:b|       |   +-- do/agent.ts    <- AgentDO (actor per agent)|       |   +-- workflows/|       |       +-- eval.ts    <- LLM judge per task|       |       +-- cluster.ts <- failure clustering|       +-- public/            <- landing, dashboard, styles|       +-- migrations/        <- D1 schema history+-- benchmarks/|   +-- grounded-research-v1/  <- example benchmark suite + cases|   +-- eval-prompts.md        <- open-source judge prompts+-- examples/                  <- runnable example agents

License

MIT — see LICENSE

Built by @virajm1shra on Cloudflare.

來源：README.md，提交 3d8bc04

工具

3

search_agents

Search public AI agents

Find listed public agents by task, capability, category, framework, model, verified evidence, or reuse configuration. Owner telemetry and controlled benchmark evidence are returned separately.

輸入參數sortlimitmodelquerylicensecategoryreusableverifiedframeworkliveCallablemaxP50LatencyMsmaxCostPerRunUsd

get_agent

Get a public agent

Get one public agent's recipe, public capability manifest, coarse invocation status, owner telemetry, and verified benchmark submissions.

輸入參數handle必填

list_benchmarks

List verified benchmark contracts

List public, versioned benchmark contracts and only their trusted-runner-verified submissions.

無需輸入參數。

版本歷史

1
  1. v0.1.0最新Sep 16, 2026

評分

已滿足適用檢查項的 100%

該評分衡量此收錄條目的完整程度，依據註冊中心條目、端點與公開來源計算得出。它不是對伺服器本身的評價，也不表示伺服器安全或目前可用：可連通性是帶日期的檢查結果，評分也不包含安裝量或使用者評價。

  • 近 30 天內可連通必需25 分
  • 已宣告認證方式必需10 分
  • 已列出工具必需15 分
  • 描述必需15 分
  • 圖示10 分
  • 文件15 分
  • 授權條款10 分