Module 09: Agentic Development And ADLC
Use agents to accelerate engineering without losing understanding, safety, or ownership.
The Mental Model
Section titled “The Mental Model”ADLC is SDLC with delegated machine assistance at each stage.
human intent -> context -> agent task -> output -> human review -> verificationThe agent can produce work. The human must produce judgment.
Analogy
Section titled “Analogy”An agent is like a very fast junior engineer who has read a lot, forgets details, sometimes invents APIs, and never carries production responsibility. Treat it as useful, not authoritative.
Agent Roles
Section titled “Agent Roles”| Role | Good For | Risk |
|---|---|---|
| Explorer | Mapping repos and finding files | May miss runtime behavior |
| Planner | Breaking down work | May ignore constraints |
| Worker | Bounded implementation | May over-edit |
| Reviewer | Finding bugs and gaps | May produce false positives |
| Operator | Summarizing logs and commands | May misread environment |
| Documenter | Drafting runbooks and handoffs | May sound confident but vague |
Visual: ADLC Control Loop
Section titled “Visual: ADLC Control Loop”agentic development control loop
GoalHuman defines outcome and constraints.
ContextFiles, commands, contracts, risks.
TaskBounded agent assignment.
OutputPatch, map, test plan, or summary.
ReviewHuman inspects diff and assumptions.
VerifyTests and runtime proof.
Accept or reviseIntegrate, reject, or retask.
ObserveDocument behavior after change.
Context Pack
Section titled “Context Pack”Every serious agent task should include:
- Goal.
- Current behavior.
- Desired behavior.
- Relevant files.
- Constraints.
- Out of scope.
- Verification commands.
- Expected response format.
Bad Agent Use
Section titled “Bad Agent Use”- “Build the whole app.”
- “Fix all bugs.”
- “Make this production ready.”
- “Deploy this.”
- “Refactor everything.”
These are vague and dangerous.
Good Agent Use
Section titled “Good Agent Use”- “Inspect the ticket creation path and return files, data flow, tests, and risks. Do not edit files.”
- “Add server-side validation for missing ticket title in
routes/tickets.ts. Add tests. Do not touch UI files.” - “Review this diff for auth bypass, duplicate writes, missing tests, and behavior regressions.”
Verification Standards
Section titled “Verification Standards”Agent output is not accepted until:
- Diff is understood.
- Tests run.
- Runtime behavior is checked if user-facing.
- Security/data assumptions are reviewed.
- Docs/contracts are updated if behavior changed.
- Remaining risks are written down.
Questions
Section titled “Questions”- What context does the agent need?
- What should the agent not touch?
- What would make the output unacceptable?
- What tests prove the output?
- What claims need manual verification?
- What did the agent miss?
- Create
portfolio/09-agentic-development-adlc/agentic-log.md. - Write a context pack for a feature.
- Ask an agent to map the relevant code.
- Ask an agent for a plan.
- Edit the plan.
- Implement one bounded slice.
- Ask an agent to review the diff.
- Run verification.
- Document accepted and rejected agent output.
Done When
Section titled “Done When”- Agent tasks are specific.
- Outputs are reviewed, not trusted.
- Verification evidence exists.
- The learner can explain every accepted change.