Introduction to governed agents
Model an agent as a bounded decision-and-action system with explicit authority, tools, memory, and evidence.
Before you start
Module prerequisites: SAI-220, SAI-240, SAI-250. Check the exercise-specific tools and environment before running commands.
Module outcomes: Define an allowlisted tool registry; Design human approval and confinement boundaries; Produce verifiable run receipts for agent actions. This lesson contributes to that sequence.
Agents cross the decision-action boundary
An agent observes context, plans or selects steps, invokes tools, changes external state, uses memory, and adapts to results. Governance must constrain the complete loop, not only the model prompt.
Define the agent's purpose, prohibited goals, initiating actors, affected resources, data, tools, side effects, autonomy, duration, budgets, memory, external dependencies, and stop conditions.
Principles
- Treat model output as an untrusted proposal, not authorization.
- Allowlist tools with typed contracts and bounded effects.
- Preserve actor, agent, workload, and tool identity.
- Authorize each material action at execution time.
- Require approval where consequence or uncertainty demands it.
- Isolate execution and cap time, spend, steps, data, and network reach.
- Produce attributable run receipts and test recovery.
Module outcome
You will create an authority boundary, tool registry, approval and confinement policy, run-receipt schema, and evaluation/incident plan.
The Basic Local Coding Agent Lab is an available introductory prototype. Begin there with Python, Ollama and Docker; the advanced module prerequisites apply to the governed-system design sequence, not to that basic exercise.
SAI-280 remains an advanced outline. Its full executable lab and practical assessment await reproducibility and safety validation. A single-file patch demonstration does not establish a governed production agent.
Module worked example
A reviewer approves a proposal, then another process changes a selected source file.
- Apply the control
- Bind approval to the displayed diff and original source hashes, then recheck all targets before writing.
- Expected result
- The stale proposal is rejected without applying an earlier file in the same change set.
- Troubleshooting
- Distinguish rejection, application failure and uncertain partial state. Reconcile side effects before retrying.
Practice and completion check
Choose one control from this lesson. Identify its actor, target, enforcement point and expected result. Describe an allowed case and a denied or failed case. For a runnable exercise, retain observed output and separate it from expected output. Record unresolved limitations before marking the lesson complete.
Your lesson progress
Completion is self-reported learning progress; it does not certify practical competence.
Status: Not marked
Loading your learning record…