Governed Agent Recipes
Four complete, runnable Python agents — each one shipping with identity, per-call policy, and an audit chain from line one, not bolted on later. Copy the file, run it, watch every tool call land in your audit log. Around 200 lines each.
New to the concepts? Start with runtime authorization and delegation chains — every recipe uses both.
Posts in this series
Each recipe is a different governance pattern: delegation, tool-output redaction, deep chains, and human-in-the-loop approval.
-
1.
Build a governed GitHub PR reviewer in Python (with subagent delegation)A Python AI agent that reviews pull requests, spawns a security-scanner and a test-runner as scope-narrowed subagents, and ships an audit chain back to the human reviewer. Full working code. ~210 lines.
-
2.
Build a governed SQL agent that scrubs PII from query results (Python, runnable)A Python AI agent that runs natural-language queries against a Postgres warehouse, with tool-output PII scrubbing happening at the governance layer — not in agent code. The agent never sees raw customer data. ~180 lines.
-
3.
Build a governed multi-step research agent (delegation chain across 4 hops)A Python research agent that decomposes a question, spawns parallel search subagents, then a synthesizer subagent — with a 4-deep delegation chain that traces every tool call back to the human asker. ~230 lines.
-
4.
Build a governed customer-support email triage agent (with human-in-the-loop on sensitive sends)A Python agent that classifies incoming support email, drafts replies, and asks for human approval before any reply that mentions refunds, account closure, or escalations is sent. The 'ask' decision flow, in working code. ~190 lines.
Curious what happens when the builder is an agent? We let a fresh Claude agent build a governed agent with no vendor steer — it picked Microsoft, and we published the whole discovery path. For the SDKs the recipes use, see the Anthropic Agent SDK integration or the governance SDKs repo.