Step 1: Create an Account
Visit seks-broker.stcredzero.workers.dev and sign up with your email.
Step 2: Add Your API Keys
Go to Secrets and add your credentials:
| Name | Value | Provider |
|---|---|---|
| OPENAI_API_KEY | sk-… | OpenAI |
| ANTHROPIC_API_KEY | sk-ant-… | Anthropic |
Keys are stored in Cloudflare D1. Application-level encryption at rest is coming soon.
Step 3: Create an Agent
Go to Agents and create one. Give it a name like “my-assistant”.
Step 4: Generate Proxy Tokens
For each provider you want to use, click Generate Token. You’ll get something like:
seks_openai_abc123def456...
This is your fake token. It’s not an OpenAI key — it’s a reference that only works through our broker.
Step 5: Update Your Agent Code
Change the base URL to point to our broker:
Python (OpenAI)
from openai import OpenAI
client = OpenAI(
api_key="seks_openai_abc123...", # Your fake token
base_url="https://seks-broker.stcredzero.workers.dev/api/openai"
)
response = client.chat.completions.create(
model="gpt-4",
messages=[{"role": "user", "content": "Hello!"}]
)
Python (Anthropic)
from anthropic import Anthropic
client = Anthropic(
api_key="seks_anthropic_xyz789...", # Your fake token
base_url="https://seks-broker.stcredzero.workers.dev/api/anthropic"
)
message = client.messages.create(
model="claude-3-opus-20240229",
max_tokens=1024,
messages=[{"role": "user", "content": "Hello!"}]
)
curl (GitHub)
curl https://seks-broker.stcredzero.workers.dev/api/github/user \
-H "Authorization: Bearer seks_github_..."
That’s It!
Your agent now uses API credentials without ever seeing them. If the agent is compromised:
- Attacker gets
seks_openai_abc123— useless outside the broker - Attacker does NOT get
sk-...— your real key stays safe
Next Steps
- Multiple agents: Create separate agents with different access
- Activity logs: Coming soon — audit logging is on the roadmap
- Token rotation: Coming soon — token regeneration is on the roadmap
Questions? Check the FAQ or reach out to the team.