Practical checklists

Before coding

• Verifiable objective defined.
• Verifiable sources identified.
• Included and excluded scope explicit.
• Risks known.
• Useful skills selected.
• Human validation requested if needed.

Lot plan

• Lot ID.
• Short context.
• Assumptions.
• Candidate files.
• Planned checks.
• User E2E tests.
• Stop conditions.
• Expected status after execution.

Before closure

• Diff reread.
• Tests executed or impossibility documented.
• Contracts validated.
• SR memory updated.
• User E2E tests listed.
• Next step clear.
• Decision done, user_testing, repair or blocked.

SR Agent Method

• Exact agent role defined.
• Authorized data listed.
• Forbidden actions explicit.
• Autonomy level defined.
• Output schema and typed model defined.
• Strict runtime validation.
• Invalid output policy.
• Non-sensitive logs.
• Human validation for critical actions.