
Security News
Anthropic Identifies Biased Reasoning and Recklessness as Drivers of Claude’s PyPI Attack
Anthropic found biased reasoning and recklessness drove Claude Mythos 5 to publish malware on PyPI and compromise a security vendor.
Seracade is a drop-in OpenAI-compatible routing proxy for AI agent teams. Six named capabilities: Call (every request, addressable and replayable), Step (sub-Call routing context inside agent trajectories), Quality Score (calibrated, version-stamped quali
Route your AI agent's LLM calls to the cheapest model that meets your quality bar — automatically. Seracade audits your real traffic, proves savings with empirical scoring, then routes every call to the Pareto-optimal model for its task type.
Free audit. 15% of per-call savings after $500/month threshold. Your API keys, never stored.
Recommended: Set your API key as an environment variable so it is never passed inline through tool calls.
export SERACADE_API_KEY=your-provider-api-key
Add to claude_desktop_config.json:
{
"mcpServers": {
"seracade": {
"command": "npx",
"args": ["-y", "seracade"],
"env": {
"SERACADE_API_KEY": "your-provider-api-key"
}
}
}
}
claude mcp add seracade -- npx -y seracade
from crewai import Agent
from crewai_tools import MCPServerAdapter
import os
mcp_adapter = MCPServerAdapter(
{"command": "npx", "args": ["-y", "seracade"]},
env={"SERACADE_API_KEY": os.environ["SERACADE_API_KEY"]}
)
agent = Agent(
role="LLM Cost Optimizer",
tools=mcp_adapter.tools
)
from langchain_mcp_adapters.client import MultiServerMCPClient
import os
client = MultiServerMCPClient({
"seracade": {
"command": "npx",
"args": ["-y", "seracade"],
"env": {"SERACADE_API_KEY": os.environ["SERACADE_API_KEY"]},
"transport": "stdio",
}
})
tools = await client.get_tools()
import { experimental_createMCPClient } from "ai";
import { Experimental_StdioMCPTransport } from "ai/mcp-stdio";
const mcp = await experimental_createMCPClient({
transport: new Experimental_StdioMCPTransport({
command: "npx",
args: ["-y", "seracade"],
env: { SERACADE_API_KEY: process.env.SERACADE_API_KEY! },
}),
});
const tools = await mcp.tools();
Step 1 — Audit. Call seracade with action=start_audit. Set OPENAI_BASE_URL=https://seracade.com/v1. Your API calls flow through unchanged — zero latency impact, your keys, your data. At 50 calls, the audit triggers automatically.
Step 2 — Review. Call seracade with action=get_status to check progress. When the report is ready, the tool returns a URL. Share it with a human decision maker before activating routing.
Step 3 — Activate. Call seracade with action=activate_routing. Every subsequent call routes to the optimal model on the efficient frontier for its task type.
Step 4 — Route. Use seracade_route in place of your direct LLM calls. Seracade selects the cheapest model that meets your quality bar and returns the full completion with per-call savings.
Step 5 — Track. Call seracade with action=check_savings to see your monthly savings. Free until the savings Seracade generates for you exceed $500/month — then 15% of the per-call price difference. You keep 85%.
seracade_routeRoute an LLM completion to the cheapest model that meets your quality bar. Requires routing to be active — call seracade with action=activate_routing first.
| Parameter | Type | Required | Description |
|---|---|---|---|
model | string | Yes | The model you would normally use (e.g. gpt-4o, claude-sonnet-4-6) |
messages | array | Yes | OpenAI-format messages array |
temperature | number | No | Sampling temperature |
max_tokens | integer | No | Max output tokens |
api_key | string | No | Your provider API key. Prefer SERACADE_API_KEY env var. |
Returns: full completion, routed model, task type, and estimated savings per call.
seracadeManage your Seracade account. Action-dispatched.
| Parameter | Type | Required | Description |
|---|---|---|---|
action | string | Yes | One of: start_audit, get_status, activate_routing, check_savings, set_budget |
api_key | string | No | Your provider API key. Prefer SERACADE_API_KEY env var. |
email | string | No | Email for report delivery (start_audit only) |
strategy | string | No | Routing strategy: balanced, max_savings, conservative (activate_routing only, default: balanced) |
agent_name | string | No | Agent identifier (set_budget only) |
monthly_budget_usd | number | No | Monthly spend cap in USD (set_budget only) |
action_on_exceed | string | No | block, downgrade, or alert when budget exceeded (set_budget only, default: alert) |
Connect directly without npx — for hosts that support streamable HTTP transport:
{
"mcpServers": {
"seracade": { "url": "https://seracade.com/mcp" }
}
}
Discovery endpoint: https://seracade.com/.well-known/mcp.json
FAQs
Seracade is a drop-in OpenAI-compatible routing proxy for AI agent teams. Six named capabilities: Call (every request, addressable and replayable), Step (sub-Call routing context inside agent trajectories), Quality Score (calibrated, version-stamped quali
The npm package seracade receives a total of 20 weekly downloads. As such, seracade popularity was classified as not popular.
We found that seracade demonstrated a healthy version release cadence and project activity because the last version was released less than a year ago. It has 1 open source maintainer collaborating on the project.

Security News
Anthropic found biased reasoning and recklessness drove Claude Mythos 5 to publish malware on PyPI and compromise a security vendor.

Research
/Security News
Malicious Chrome and Firefox extensions target Axiom Trade and Padre users, stealing session tokens and wallet data.

Security News
GPT-6 Astra hits 100% on ExploitBench and finds zero-days autonomously, while independent tests reveal scope violations and monitoring gaps.