OpenAI Agents SDK: guard email with CAIN

1 Get a key

Give each agent its own key in production (console → Agent keys): decisions, trust history and revocation are then per agent.

2 Install

pip install "cainstudio[openai-agents] @ https://cainstudio.online/cainstudio-0.4.1-py3-none-any.whl#sha256=1bea5cf49aeeb2775acec58933fc39b008a1c84bfd7c9690c54723ed707af689"
export CAIN_API_KEY=...

cainstudio 0.4.1, pinned by URL and SHA-256 so pip installs exactly the published wheel (the package is not on PyPI yet). Framework dependencies come from PyPI as usual.

3 Put CAIN in front of function tools

protect() attaches a tool input guardrail to each tool; the SDK runs it immediately before the tool executes.

from agents import Agent, Runner, function_tool
from cainstudio.integrations.openai_agents import protect

@function_tool
def send_email(to: str, subject: str, body: str) -> str:
    """Email: an agent that sends email."""
    ...  # your existing code, unchanged

agent = Agent(name='email-agent', instructions="...",
              tools=protect([send_email], agent_id='email-agent'))
result = Runner.run_sync(agent, "...")

On a refusal the agent gets reject_content with the reason (on_block="raise" trips a tripwire and halts the run instead). The tool body never runs: not on a refusal, not on a hold, not when CAIN cannot be reached.

4 See what CAIN decides for this exact call

This is the request the integration sends when the model calls send_email with these arguments. record: false evaluates it without writing an evidence record. A new agent has no trust history, so expect REQUIRE_APPROVAL at first; destinations outside your egress allowlist are refused by the egress stage.

5 Write the rule, and test it before you save it

Mail inside your company runs; mail to anyone outside waits for a person. Recipients are also destinations: until you allow a destination under egress rules, the egress stage refuses it on its own.

The panel sends the draft rule to the rule tester: it says which rule would decide the call and saves nothing. Swap in the safe arguments below to see it pass through. matches is a case-insensitive glob (*, ?), not a regular expression; conditions on one rule must all hold; the first matching rule by priority wins.

Safe arguments: {"to": "ops@yourcompany.com", "subject": "Invoice", "body": "..."}

When it does what you want, save it from the console or with POST /fabric/tool-rules:

curl -sS -X POST https://cainstudio.online/fabric/tool-rules \
  -H "X-API-Key: $CAIN_API_KEY" -H 'Content-Type: application/json' \
  -d '{"name": "email-guard", "effect": "require_approval", "priority": 10, "match_path": "/tools/send_email", "conditions": [{"field": "to", "op": "not_contains", "value": "@yourcompany.com"}]}'

6 Handle the calls that wait for a person

A held call is not executed. It appears in the approval queue with its exact arguments; an approval lets that one call run once. Adapters can wait instead of returning: wait_for_approval=120.

7 Check it before you ship

cainstudio redteam --agent email-agent   # dry-runs dangerous calls; exit 2 if any would run
cainstudio decisions                 # what was decided, and why
cainstudio explain <decision_id>

Every recorded decision is signed; Cain().verify(decision) checks its quorum certificate on your machine (Python SDK).