OpenTelemetry: guard sql 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[otel] @ 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 your existing tracing

instrument() wraps each decision in an execute_tool span under the current span, with gen_ai.* and cain.* attributes. Arguments are not recorded unless you pass record_arguments=True.

from cainstudio import Cain
from cainstudio.integrations.otel import instrument
from cainstudio.integrations.universal import guard_tools

def run_sql(query: str) -> str:
    """SQL: an agent with database access."""
    ...  # your existing code, unchanged

cain = instrument(Cain())   # every decision becomes an execute_tool span
tools = guard_tools([run_sql], client=cain, agent_id='sql-agent')

On a refusal the agent gets whatever the guard in front of it returns; the span carries cain.verdict. 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 run_sql 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

Reads run; dropping a table is refused outright. Text is compared after normalising case, whitespace, SQL comments and look-alike characters, so DROP/**/table and full-width letters are caught too. Add one rule per statement you never want (truncate, alter, grant).

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: {"query": "SELECT id FROM customers LIMIT 10"}

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": "sql-guard", "effect": "deny", "priority": 10, "match_path": "/tools/run_sql", "conditions": [{"field": "query", "op": "matches", "value": "*drop table*"}]}'

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 sql-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).