Small Group Tutorials

Here to help students catch up, keep up, and move ahead. Book a consultation here.

How to Perform in the new G2 SEC Examinations | Learner’s Guide Vol 0172 | Handoff Contracts — Pass the Right Information From One Reasoning Step to the Next

G2 SEC examination performance is not only about solving individual steps. It is also about passing information safely from one step to the next. A correct intermediate value can become wrong when its unit is lost; a valid quote can become misleading when its speaker is lost; a sound Science observation can become an overclaim when its conditions are lost.

This one-hundred-and-seventy-second Learner’s Guide develops handoff contracts across K200 English, K210 Mathematics and K223–K225 Science: the minimum information an output must carry so the next reasoning step can use it without guessing.

Mechanism: interfaces can corrupt correct work

Each stage consumes an input and produces an output. The next stage needs more than the raw content: it may need source, unit, scope, condition, precision, time or location. A handoff contract specifies that metadata. Many examination errors are interface failures between otherwise correct steps.

Diagnosis

When a later step fails, inspect the intermediate it received. Ask whether the value/claim arrived with enough ownership information to be used correctly. If the next step had to infer the unit, source, denominator, branch or condition, the handoff was under-specified.

Smallest repair

Restore the missing metadata at the interface—label the value, attach the condition, name the source, preserve precision—then rerun only the dependent step. Do not rebuild the upstream solution if its content was already correct.

1. Question to plan

Contract: The plan should receive command, scope, source and required count—not just the topic.

Control: Write a compact task state before generating content.

For practice, write the upstream output on a separate line as if another learner must continue from it. Include only the information the next step genuinely needs. Then hide the original problem and test whether the continuation remains unambiguous.

For transfer, change the surface context while preserving the same interface. The learner should rebuild the handoff fields from the live task rather than memorise one annotation pattern.

2. Plan to paragraph

Contract: Each paragraph should receive one function, claim and evidence role.

Control: Do not hand a vague ‘write about X’ instruction to the paragraph.

For practice, write the upstream output on a separate line as if another learner must continue from it. Include only the information the next step genuinely needs. Then hide the original problem and test whether the continuation remains unambiguous.

For transfer, change the surface context while preserving the same interface. The learner should rebuild the handoff fields from the live task rather than memorise one annotation pattern.

3. Paragraph to conclusion

Contract: The conclusion should receive established claims, not reopen unsupported ideas.

Control: Only hand off what earlier paragraphs have earned.

For practice, write the upstream output on a separate line as if another learner must continue from it. Include only the information the next step genuinely needs. Then hide the original problem and test whether the continuation remains unambiguous.

For transfer, change the surface context while preserving the same interface. The learner should rebuild the handoff fields from the live task rather than memorise one annotation pattern.

4. Source to inference

Contract: Evidence should hand over source owner, wording strength and context.

Control: Inference without source metadata can overclaim.

For practice, write the upstream output on a separate line as if another learner must continue from it. Include only the information the next step genuinely needs. Then hide the original problem and test whether the continuation remains unambiguous.

For transfer, change the surface context while preserving the same interface. The learner should rebuild the handoff fields from the live task rather than memorise one annotation pattern.

5. Quote to explanation

Contract: The explanation should receive the exact feature being interpreted.

Control: Do not quote broadly and explain something else.

For practice, write the upstream output on a separate line as if another learner must continue from it. Include only the information the next step genuinely needs. Then hide the original problem and test whether the continuation remains unambiguous.

For transfer, change the surface context while preserving the same interface. The learner should rebuild the handoff fields from the live task rather than memorise one annotation pattern.

6. Example to general point

Contract: The example should hand over only the property it demonstrates.

Control: One case should not silently become a universal rule.

For practice, write the upstream output on a separate line as if another learner must continue from it. Include only the information the next step genuinely needs. Then hide the original problem and test whether the continuation remains unambiguous.

For transfer, change the surface context while preserving the same interface. The learner should rebuild the handoff fields from the live task rather than memorise one annotation pattern.

7. Reading to summary

Contract: Selected points should arrive with focus relevance and uniqueness.

Control: Strip examples only after the general point is secure.

For practice, write the upstream output on a separate line as if another learner must continue from it. Include only the information the next step genuinely needs. Then hide the original problem and test whether the continuation remains unambiguous.

For transfer, change the surface context while preserving the same interface. The learner should rebuild the handoff fields from the live task rather than memorise one annotation pattern.

8. Listening notes to answer

Contract: Notes should retain speaker, uncertainty and sequence.

Control: Anonymous fragments create false certainty.

For practice, write the upstream output on a separate line as if another learner must continue from it. Include only the information the next step genuinely needs. Then hide the original problem and test whether the continuation remains unambiguous.

For transfer, change the surface context while preserving the same interface. The learner should rebuild the handoff fields from the live task rather than memorise one annotation pattern.

9. Oral idea to example

Contract: The example should receive a clear claim/reason to illustrate.

Control: Do not let the example choose the point after the fact.

For practice, write the upstream output on a separate line as if another learner must continue from it. Include only the information the next step genuinely needs. Then hide the original problem and test whether the continuation remains unambiguous.

For transfer, change the surface context while preserving the same interface. The learner should rebuild the handoff fields from the live task rather than memorise one annotation pattern.

10. Writing draft to edit

Contract: Editing should receive task obligations and meaning invariants.

Control: Polish must not delete required content.

For practice, write the upstream output on a separate line as if another learner must continue from it. Include only the information the next step genuinely needs. Then hide the original problem and test whether the continuation remains unambiguous.

For transfer, change the surface context while preserving the same interface. The learner should rebuild the handoff fields from the live task rather than memorise one annotation pattern.

11. Math story to variables

Contract: Each symbol receives quantity, unit and domain.

Control: Anonymous symbols are broken handoffs.

For practice, write the upstream output on a separate line as if another learner must continue from it. Include only the information the next step genuinely needs. Then hide the original problem and test whether the continuation remains unambiguous.

For transfer, change the surface context while preserving the same interface. The learner should rebuild the handoff fields from the live task rather than memorise one annotation pattern.

12. Variables to equation

Contract: Equation receives relationships and signs, not just values.

Control: Term provenance must survive compression.

For practice, write the upstream output on a separate line as if another learner must continue from it. Include only the information the next step genuinely needs. Then hide the original problem and test whether the continuation remains unambiguous.

For transfer, change the surface context while preserving the same interface. The learner should rebuild the handoff fields from the live task rather than memorise one annotation pattern.

13. Equation to solver

Contract: Solver receives domain and branch restrictions.

Control: Algebra alone cannot decide contextual validity.

For practice, write the upstream output on a separate line as if another learner must continue from it. Include only the information the next step genuinely needs. Then hide the original problem and test whether the continuation remains unambiguous.

For transfer, change the surface context while preserving the same interface. The learner should rebuild the handoff fields from the live task rather than memorise one annotation pattern.

14. Solver to root filter

Contract: Candidate roots arrive with original equation and context.

Control: Filtering without provenance can reject/accept wrongly.

For practice, write the upstream output on a separate line as if another learner must continue from it. Include only the information the next step genuinely needs. Then hide the original problem and test whether the continuation remains unambiguous.

For transfer, change the surface context while preserving the same interface. The learner should rebuild the handoff fields from the live task rather than memorise one annotation pattern.

15. Intermediate to next formula

Contract: The value arrives with label, unit and precision.

Control: Do not pass bare calculator numbers.

For practice, write the upstream output on a separate line as if another learner must continue from it. Include only the information the next step genuinely needs. Then hide the original problem and test whether the continuation remains unambiguous.

For transfer, change the surface context while preserving the same interface. The learner should rebuild the handoff fields from the live task rather than memorise one annotation pattern.

16. Graph to calculation

Contract: Read-off arrives with series, coordinates, scale and unit.

Control: A copied number without axis ownership is unsafe.

For practice, write the upstream output on a separate line as if another learner must continue from it. Include only the information the next step genuinely needs. Then hide the original problem and test whether the continuation remains unambiguous.

For transfer, change the surface context while preserving the same interface. The learner should rebuild the handoff fields from the live task rather than memorise one annotation pattern.

17. Table to calculation

Contract: Value arrives with row, column, heading and unit.

Control: Traceability prevents wrong-cell use.

For practice, write the upstream output on a separate line as if another learner must continue from it. Include only the information the next step genuinely needs. Then hide the original problem and test whether the continuation remains unambiguous.

For transfer, change the surface context while preserving the same interface. The learner should rebuild the handoff fields from the live task rather than memorise one annotation pattern.

18. Ratio to percentage

Contract: The fraction arrives with named base.

Control: Percentage conversion must not lose denominator meaning.

For practice, write the upstream output on a separate line as if another learner must continue from it. Include only the information the next step genuinely needs. Then hide the original problem and test whether the continuation remains unambiguous.

For transfer, change the surface context while preserving the same interface. The learner should rebuild the handoff fields from the live task rather than memorise one annotation pattern.

19. Rate to total

Contract: Rate arrives with denominator unit and active duration.

Control: Multiply only by matching exposure.

For practice, write the upstream output on a separate line as if another learner must continue from it. Include only the information the next step genuinely needs. Then hide the original problem and test whether the continuation remains unambiguous.

For transfer, change the surface context while preserving the same interface. The learner should rebuild the handoff fields from the live task rather than memorise one annotation pattern.

20. Total to normalised rate

Contract: Raw total arrives with population/base.

Control: Normalisation needs ownership of both numerator and denominator.

For practice, write the upstream output on a separate line as if another learner must continue from it. Include only the information the next step genuinely needs. Then hide the original problem and test whether the continuation remains unambiguous.

For transfer, change the surface context while preserving the same interface. The learner should rebuild the handoff fields from the live task rather than memorise one annotation pattern.

21. Geometry diagram to theorem

Contract: Theorem receives verified conditions and correspondence.

Control: Visual appearance is not a valid handoff.

For practice, write the upstream output on a separate line as if another learner must continue from it. Include only the information the next step genuinely needs. Then hide the original problem and test whether the continuation remains unambiguous.

For transfer, change the surface context while preserving the same interface. The learner should rebuild the handoff fields from the live task rather than memorise one annotation pattern.

22. Theorem to calculation

Contract: Derived equality/ratio arrives with reason.

Control: Do not treat theorem output as given data.

For practice, write the upstream output on a separate line as if another learner must continue from it. Include only the information the next step genuinely needs. Then hide the original problem and test whether the continuation remains unambiguous.

For transfer, change the surface context while preserving the same interface. The learner should rebuild the handoff fields from the live task rather than memorise one annotation pattern.

23. Case split to branch

Contract: Each branch receives its activating condition.

Control: Formula and condition travel together.

For practice, write the upstream output on a separate line as if another learner must continue from it. Include only the information the next step genuinely needs. Then hide the original problem and test whether the continuation remains unambiguous.

For transfer, change the surface context while preserving the same interface. The learner should rebuild the handoff fields from the live task rather than memorise one annotation pattern.

24. Branch to recombination

Contract: Component answer arrives with sign/weight and scope.

Control: Correct parts can be recombined wrongly.

For practice, write the upstream output on a separate line as if another learner must continue from it. Include only the information the next step genuinely needs. Then hide the original problem and test whether the continuation remains unambiguous.

For transfer, change the surface context while preserving the same interface. The learner should rebuild the handoff fields from the live task rather than memorise one annotation pattern.

25. Exact to decimal

Contract: Exact value arrives with requested accuracy and rounding stage.

Control: Do not round before the interface requires it.

For practice, write the upstream output on a separate line as if another learner must continue from it. Include only the information the next step genuinely needs. Then hide the original problem and test whether the continuation remains unambiguous.

For transfer, change the surface context while preserving the same interface. The learner should rebuild the handoff fields from the live task rather than memorise one annotation pattern.

26. Working to final answer

Contract: Final response receives value, unit, label and question number.

Control: Transfer is its own interface.

For practice, write the upstream output on a separate line as if another learner must continue from it. Include only the information the next step genuinely needs. Then hide the original problem and test whether the continuation remains unambiguous.

For transfer, change the surface context while preserving the same interface. The learner should rebuild the handoff fields from the live task rather than memorise one annotation pattern.

27. Science question to design

Contract: Design receives target causal/associational question and variables.

Control: Method should answer the question, not merely look scientific.

For practice, write the upstream output on a separate line as if another learner must continue from it. Include only the information the next step genuinely needs. Then hide the original problem and test whether the continuation remains unambiguous.

For transfer, change the surface context while preserving the same interface. The learner should rebuild the handoff fields from the live task rather than memorise one annotation pattern.

28. Design to apparatus

Contract: Apparatus receives required range, sensitivity and timing.

Control: Instrument choice follows measurement need.

For practice, write the upstream output on a separate line as if another learner must continue from it. Include only the information the next step genuinely needs. Then hide the original problem and test whether the continuation remains unambiguous.

For transfer, change the surface context while preserving the same interface. The learner should rebuild the handoff fields from the live task rather than memorise one annotation pattern.

29. Treatment plan to manipulation

Contract: Manipulation receives intended independent-variable state.

Control: Verify actual system exposure.

For practice, write the upstream output on a separate line as if another learner must continue from it. Include only the information the next step genuinely needs. Then hide the original problem and test whether the continuation remains unambiguous.

For transfer, change the surface context while preserving the same interface. The learner should rebuild the handoff fields from the live task rather than memorise one annotation pattern.

30. Manipulation to outcome

Contract: Outcome interpretation receives evidence that treatment contrast existed.

Control: Otherwise null/positive result is ambiguous.

For practice, write the upstream output on a separate line as if another learner must continue from it. Include only the information the next step genuinely needs. Then hide the original problem and test whether the continuation remains unambiguous.

For transfer, change the surface context while preserving the same interface. The learner should rebuild the handoff fields from the live task rather than memorise one annotation pattern.

31. Control to conclusion

Contract: Conclusion receives which alternative explanation the control addresses.

Control: ‘Fair test’ is insufficient metadata.

For practice, write the upstream output on a separate line as if another learner must continue from it. Include only the information the next step genuinely needs. Then hide the original problem and test whether the continuation remains unambiguous.

For transfer, change the surface context while preserving the same interface. The learner should rebuild the handoff fields from the live task rather than memorise one annotation pattern.

32. Calibration to measurement

Contract: Measurement receives valid range, mapping and units.

Control: Do not use calibration outside its range.

For practice, write the upstream output on a separate line as if another learner must continue from it. Include only the information the next step genuinely needs. Then hide the original problem and test whether the continuation remains unambiguous.

For transfer, change the surface context while preserving the same interface. The learner should rebuild the handoff fields from the live task rather than memorise one annotation pattern.

33. Sensor to dataset

Contract: Data receive timestamp, location, unit and calibration state.

Control: Numbers without provenance cannot support strong inference.

For practice, write the upstream output on a separate line as if another learner must continue from it. Include only the information the next step genuinely needs. Then hide the original problem and test whether the continuation remains unambiguous.

For transfer, change the surface context while preserving the same interface. The learner should rebuild the handoff fields from the live task rather than memorise one annotation pattern.

34. Sample to population

Contract: Generalisation receives sampling frame, place, time and condition.

Control: Representativeness metadata travels with the sample.

For practice, write the upstream output on a separate line as if another learner must continue from it. Include only the information the next step genuinely needs. Then hide the original problem and test whether the continuation remains unambiguous.

For transfer, change the surface context while preserving the same interface. The learner should rebuild the handoff fields from the live task rather than memorise one annotation pattern.

35. Observation to inference

Contract: Inference receives the exact observation and uncertainty.

Control: Do not merge layers.

For practice, write the upstream output on a separate line as if another learner must continue from it. Include only the information the next step genuinely needs. Then hide the original problem and test whether the continuation remains unambiguous.

For transfer, change the surface context while preserving the same interface. The learner should rebuild the handoff fields from the live task rather than memorise one annotation pattern.

36. Inference to mechanism

Contract: Mechanism receives causal/design support and competing explanations.

Control: A plausible inference is not automatically mechanism proof.

For practice, write the upstream output on a separate line as if another learner must continue from it. Include only the information the next step genuinely needs. Then hide the original problem and test whether the continuation remains unambiguous.

For transfer, change the surface context while preserving the same interface. The learner should rebuild the handoff fields from the live task rather than memorise one annotation pattern.

37. Mechanism to prediction

Contract: Prediction receives boundary conditions and range.

Control: Models should not hand off unlimited validity.

For practice, write the upstream output on a separate line as if another learner must continue from it. Include only the information the next step genuinely needs. Then hide the original problem and test whether the continuation remains unambiguous.

For transfer, change the surface context while preserving the same interface. The learner should rebuild the handoff fields from the live task rather than memorise one annotation pattern.

38. Prediction to test

Contract: Test receives discriminating outcome and measurement schedule.

Control: A test must sample where models disagree.

For practice, write the upstream output on a separate line as if another learner must continue from it. Include only the information the next step genuinely needs. Then hide the original problem and test whether the continuation remains unambiguous.

For transfer, change the surface context while preserving the same interface. The learner should rebuild the handoff fields from the live task rather than memorise one annotation pattern.

39. Test to revision

Contract: Revision receives which assumption failed.

Control: Do not rewrite unrelated model components.

For practice, write the upstream output on a separate line as if another learner must continue from it. Include only the information the next step genuinely needs. Then hide the original problem and test whether the continuation remains unambiguous.

For transfer, change the surface context while preserving the same interface. The learner should rebuild the handoff fields from the live task rather than memorise one annotation pattern.

40. Baseline to comparison

Contract: Comparison receives baseline stability and timing.

Control: A drifting reference is a bad handoff.

For practice, write the upstream output on a separate line as if another learner must continue from it. Include only the information the next step genuinely needs. Then hide the original problem and test whether the continuation remains unambiguous.

For transfer, change the surface context while preserving the same interface. The learner should rebuild the handoff fields from the live task rather than memorise one annotation pattern.

41. Compartment to whole system

Contract: Whole-system claim receives compartment size, flows and boundary assumptions.

Control: Local state cannot be handed off as global automatically.

For practice, write the upstream output on a separate line as if another learner must continue from it. Include only the information the next step genuinely needs. Then hide the original problem and test whether the continuation remains unambiguous.

For transfer, change the surface context while preserving the same interface. The learner should rebuild the handoff fields from the live task rather than memorise one annotation pattern.

42. Spatial sample to map

Contract: Map receives location coordinates/labels and measurement conditions.

Control: Location ownership matters.

For practice, write the upstream output on a separate line as if another learner must continue from it. Include only the information the next step genuinely needs. Then hide the original problem and test whether the continuation remains unambiguous.

For transfer, change the surface context while preserving the same interface. The learner should rebuild the handoff fields from the live task rather than memorise one annotation pattern.

43. Time sample to curve

Contract: Curve receives actual timestamps and sampling cadence.

Control: Do not imply unmeasured shape.

For practice, write the upstream output on a separate line as if another learner must continue from it. Include only the information the next step genuinely needs. Then hide the original problem and test whether the continuation remains unambiguous.

For transfer, change the surface context while preserving the same interface. The learner should rebuild the handoff fields from the live task rather than memorise one annotation pattern.

44. Replicates to mean

Contract: Mean receives count, spread and exclusion decisions.

Control: Average should not erase variability provenance.

For practice, write the upstream output on a separate line as if another learner must continue from it. Include only the information the next step genuinely needs. Then hide the original problem and test whether the continuation remains unambiguous.

For transfer, change the surface context while preserving the same interface. The learner should rebuild the handoff fields from the live task rather than memorise one annotation pattern.

45. Anomaly to diagnosis

Contract: Diagnosis receives raw anomalous value and method context.

Control: Do not delete before handoff.

For practice, write the upstream output on a separate line as if another learner must continue from it. Include only the information the next step genuinely needs. Then hide the original problem and test whether the continuation remains unambiguous.

For transfer, change the surface context while preserving the same interface. The learner should rebuild the handoff fields from the live task rather than memorise one annotation pattern.

46. Failure diagnosis to repair

Contract: Repair receives first broken node and dependency set.

Control: Fix only descendants.

For practice, write the upstream output on a separate line as if another learner must continue from it. Include only the information the next step genuinely needs. Then hide the original problem and test whether the continuation remains unambiguous.

For transfer, change the surface context while preserving the same interface. The learner should rebuild the handoff fields from the live task rather than memorise one annotation pattern.

47. Repair to verification

Contract: Verification receives expected recovered invariant.

Control: Check the repair by a different route.

For practice, write the upstream output on a separate line as if another learner must continue from it. Include only the information the next step genuinely needs. Then hide the original problem and test whether the continuation remains unambiguous.

For transfer, change the surface context while preserving the same interface. The learner should rebuild the handoff fields from the live task rather than memorise one annotation pattern.

48. Verification to closure

Contract: Closure receives proof obligations and remaining uncertainty.

Control: Stop only when live requirements are paid.

For practice, write the upstream output on a separate line as if another learner must continue from it. Include only the information the next step genuinely needs. Then hide the original problem and test whether the continuation remains unambiguous.

For transfer, change the surface context while preserving the same interface. The learner should rebuild the handoff fields from the live task rather than memorise one annotation pattern.

49. Section to section

Contract: Next section receives only the state it needs, not unresolved clutter.

Control: This is distinct from timing handoff: it is information-interface design.

For practice, write the upstream output on a separate line as if another learner must continue from it. Include only the information the next step genuinely needs. Then hide the original problem and test whether the continuation remains unambiguous.

For transfer, change the surface context while preserving the same interface. The learner should rebuild the handoff fields from the live task rather than memorise one annotation pattern.

50. Memory to retrieval

Contract: Retrieved knowledge should arrive with conditions for use.

Control: A formula remembered without applicability is incomplete state.

For practice, write the upstream output on a separate line as if another learner must continue from it. Include only the information the next step genuinely needs. Then hide the original problem and test whether the continuation remains unambiguous.

For transfer, change the surface context while preserving the same interface. The learner should rebuild the handoff fields from the live task rather than memorise one annotation pattern.

51. Teacher feedback to next attempt

Contract: Feedback should hand over one diagnosis and repair cue.

Control: Vague ‘be careful’ feedback has low transfer value.

For practice, write the upstream output on a separate line as if another learner must continue from it. Include only the information the next step genuinely needs. Then hide the original problem and test whether the continuation remains unambiguous.

For transfer, change the surface context while preserving the same interface. The learner should rebuild the handoff fields from the live task rather than memorise one annotation pattern.

52. Practice to delayed transfer

Contract: Later task receives no surface hints from the original.

Control: The skill should reconstruct itself.

For practice, write the upstream output on a separate line as if another learner must continue from it. Include only the information the next step genuinely needs. Then hide the original problem and test whether the continuation remains unambiguous.

For transfer, change the surface context while preserving the same interface. The learner should rebuild the handoff fields from the live task rather than memorise one annotation pattern.

53. Handoff checksum

Contract: Before using an intermediate, ask whether label, unit, source, scope and condition survived.

Control: A five-second checksum prevents silent corruption.

For practice, write the upstream output on a separate line as if another learner must continue from it. Include only the information the next step genuinely needs. Then hide the original problem and test whether the continuation remains unambiguous.

For transfer, change the surface context while preserving the same interface. The learner should rebuild the handoff fields from the live task rather than memorise one annotation pattern.

54. Handoff final rule

Contract: Every intermediate output should arrive at the next step with the metadata needed to use it correctly.

Control: If the next step has to guess what the value or claim means, the handoff is broken.

For practice, write the upstream output on a separate line as if another learner must continue from it. Include only the information the next step genuinely needs. Then hide the original problem and test whether the continuation remains unambiguous.

For transfer, change the surface context while preserving the same interface. The learner should rebuild the handoff fields from the live task rather than memorise one annotation pattern.

Competing diagnoses

A wrong later result can come from bad upstream reasoning or a broken handoff. Distinguish them by checking whether the intermediate content was correct before transfer. If correct content was misused after its label, source or condition disappeared, repair the interface rather than reteach the concept.

Observation versus inference

The intermediate value, quote or measurement is observable working state. The interpretation attached to it is an inference. A good handoff preserves enough provenance that the next inference can be audited.

Support versus proof

A complete handoff prevents one class of error but does not prove the upstream content correct. Interface integrity and substantive correctness are separate checks; strong work needs both.

Model limits

Not every step needs every label. Over-annotating can create cognitive load. The contract should carry the smallest state necessary for the next step to proceed without guessing. What is necessary depends on the task.

Delayed transfer

Return later with multi-stage problems where one intermediate is deliberately stripped of unit, source, condition or scope. Measure whether the learner notices the broken interface, reconstructs the missing metadata and repairs only the affected downstream work.

Internal learning links

Use the Examination Craft hub, Vol 0100 Checkpoint Compression, Vol 0168 Requirement Traceability, the relevant English, Mathematics and Science hubs, and the PSLE Learner’s Guide.

MOE and SEAB current framework

MOE states that Full Subject-Based Banding is fully implemented and that, from 2027, the Singapore-Cambridge Secondary Education Certificate (SEC) replaces the former N- and O-Level examinations, with graduating students sitting subjects at their respective G1, G2 or G3 levels. See the official MOE Full SBB / SEC announcement. For current 2027 G2 school-candidate subject codes and syllabuses, use the official SEAB G2 syllabus directory, which lists K200 English Language, K210 Mathematics and K223–K225 Science combinations.

Final rule

Do not pass bare answers between reasoning steps. Pass the smallest complete state the next step needs: meaning, ownership, unit, source, condition, scope or precision. A correct output with missing metadata is a fragile input.

G2 SEC Learner’s Guide: Open the Vol 0132–0175 hub and subject index.