What you will be able to do#
- Distinguish six obstacle classes instead of responding to every stall with another random technique.
- Use a three-stage hint ladder and stop revealing when productive work can resume.
- Record assistance truthfully and continue with a bounded next step.
Name the obstacle first#
A precise obstacle suggests a precise recovery. 'I am stuck' does not.
- Misunderstanding: the input, output, or example is still unclear.
- Missing prerequisite: a concept such as recursion frames, prefix state, or graph representation is unavailable.
- Approach selection: the baseline is known but its repeated work has not led to a valid technique.
- Implementation: the invariant is sound but code structure, syntax, or update order is failing.
- Complexity: the code works but the operation count or bound is unclear.
- Testing: a mismatch exists but the failing branch or first divergent state is unknown.
Match the remedy to the obstacle#
Rephrase a misunderstood contract. Draw a tiny state trace for a missing representation. Enumerate a correct baseline and mark its repeated work for approach selection. Restate the invariant before repairing implementation. Count operations by variable for complexity. Shrink the counterexample and inspect the first mismatch for testing.
Changing patterns at random adds motion without information. Keep the last sound statement and take one bounded step from it.
Use a three-stage hint ladder#
Level 1 - observation or representation: expose what one item or boundary should mean. Level 2 - invariant or data structure: name the state that can remove repeated work. Level 3 - transition or pseudocode: show how that state changes. Reveal a full solution only through a separate explicit action after these levels.
- Ask for the smallest level that addresses the named obstacle.
- Explain what the hint changed in your model.
- Resume without opening the next level if you can proceed.
- Record the highest assistance level used before reviewing the result.
Keep the full solution behind an explicit action#
A solution shown automatically turns the session into guided study and can erase the opportunity to locate the actual obstacle. If you choose to reveal it, compare its state definition and update order with your own attempt, then reconstruct later without claiming the current result was independent.
Record assistance without punishment#
Hints are legitimate learning tools. Record how many were used, what level they reached, and which decision changed. An assisted solution can support engagement, debugging, and later reconstruction; it does not support the same recognition evidence as an unseen independent attempt.
Ask for targeted help#
A useful request exposes what you know and the smallest uncertain decision: 'I can enumerate every candidate in quadratic time, but I am repeating the same membership scan. Could I check whether storing the processed prefix is the intended direction?'
A live interviewer may respond with a different clue or none at all. This wording is a transparent request, not a script for employer policy.
Worked example#
Recover from a repeated membership scan
The learner has a correct quadratic pair search but cannot identify the next step.
- Classify the obstacle as approach selection, not implementation.
- Name the repeated operation: scanning for a complement for every value.
- Request a level-two hint about state or data structure only if the stored prefix idea still does not follow.
- Define the map contents and check-before-update invariant before returning to code.
Outcome: The recovery preserves the correct baseline, records assistance, and resumes from a stated invariant rather than a copied solution.
Apply the idea#
For each of these stalls - unclear output, forgotten recursion base case, no optimization after a quadratic baseline, wrong map update order, uncertain space cost, failing duplicate case - name the obstacle class and one matching recovery action.
Reveal the worked answer
Unclear output is misunderstanding: restate the contract and ask a consequential question. A forgotten base case is a missing prerequisite: trace one call frame and termination. No optimization is approach selection: mark repeated work in the baseline. Wrong update order is implementation: restate what the map must contain before each step. Uncertain space is complexity: count retained state and recursion by input variable. A duplicate failure is testing: shrink it and find the first state mismatch.
Choose the next session#
Use the recovery ladder on one current attempt, record any hint that changes the approach, and reveal a full solution only by explicit choice.
Open the stuck-recovery guideSource and review record#
This original Engineering Foundry guide was checked by Codex against the recorded research scope on 2026-09-05. It has not received independent human editorial review.
Evidence IDs: SRC-DSA-LEETCODE-EXPLORE SRC-DSA-TECH-HANDBOOK-PLAN