Practical notes: Prompt Engineering for LLMs: Zero-Shot, Few-Shot, Reasoning

Use this as an operator-facing rebuild of the ideas in “Prompt Engineering for LLMs: Zero-Shot, Few-Shot, Reasoning, Decomposition & ReAct”: clear stages, ordered code slots, and recovery notes that survive a handoff. Overview works best when treated as a measurable surface. Capture one golden transcript, one failure case, and the rollback note before expanding scope. Prefer small, testable units over sprawling scripts. When a step fails, the failure should point at a single responsibility rather than a tangled pipeline.

Zero-shot prompting

For Zero-shot prompting, define the inputs, the owner of the step, and the exit criteria before changing code. Operators should be able to re-run the step from a known checkpoint without guessing hidden state. Treat this stage as a contract between inputs and validated outputs. Name the artifacts, define success checks, and refuse silent partial completion. Prefer structured outputs with schema validation over free-form prose when the next step is code or a tool call.

Summarize the following customer complaint in one sentence.
<complaint>
The package arrived three days late and the box was damaged.
</complaint>

Clear task specification

For Clear task specification, define the inputs, the owner of the step, and the exit criteria before changing code. Operators should be able to re-run the step from a known checkpoint without guessing hidden state. Record timings and token or query cost next to functional results. Cost visibility early prevents surprise bills when the path moves from demo to shared environments. Prefer structured outputs with schema validation over free-form prose when the next step is code or a tool call.

Task:
Summarize the customer complaint.

Source:
<complaint>
The package arrived three days late and the box was damaged.
</complaint>

Constraints:
- Use only the information in the complaint.
- Do not infer the cause of the damage.
- Use one sentence.

Output:
Return only the summary.

Explicit objectives

For Explicit objectives, define the inputs, the owner of the step, and the exit criteria before changing code. Operators should be able to re-run the step from a known checkpoint without guessing hidden state. Keep configuration outside application code. Environment files, secret stores, and feature flags belong in one place operators can audit without reading the whole graph. Prefer structured outputs with schema validation over free-form prose when the next step is code or a tool call. For Explicit objectives, define the inputs, the owner of the step, and the exit criteria before changing code. Operators should be able to re-run the step from a known checkpoint without guessing hidden state. Prefer small, testable units over sprawling scripts. When a step fails, the failure should point at a single responsibility rather than a tangled pipeline.

Constraints are part of the task

When working through Constraints are part of the task, write down the contract first: required inputs, success signal, and what happens on partial failure. That checklist keeps later code changes honest. Treat this stage as a contract between inputs and validated outputs. Name the artifacts, define success checks, and refuse silent partial completion. Cache stable system instructions and tool schemas. Re-sending identical preamble is a common source of burn.

When zero-shot prompting works

When working through When zero-shot prompting works, write down the contract first: required inputs, success signal, and what happens on partial failure. That checklist keeps later code changes honest. Record timings and token or query cost next to functional results. Cost visibility early prevents surprise bills when the path moves from demo to shared environments. Cache stable system instructions and tool schemas. Re-sending identical preamble is a common source of burn.

Zero-shot failure modes

When working through Zero-shot failure modes, write down the contract first: required inputs, success signal, and what happens on partial failure. That checklist keeps later code changes honest. Keep configuration outside application code. Environment files, secret stores, and feature flags belong in one place operators can audit without reading the whole graph. Cache stable system instructions and tool schemas. Re-sending identical preamble is a common source of burn. When working through Zero-shot failure modes, write down the contract first: required inputs, success signal, and what happens on partial failure. That checklist keeps later code changes honest. Prefer small, testable units over sprawling scripts. When a step fails, the failure should point at a single responsibility rather than a tangled pipeline.

Few-shot prompting

Few-shot prompting works best when treated as a measurable surface. Capture one golden transcript, one failure case, and the rollback note before expanding scope. Treat this stage as a contract between inputs and validated outputs. Name the artifacts, define success checks, and refuse silent partial completion. Budget tokens per turn and per session. Agentic tools expand context aggressively; hard caps keep demos from becoming surprise invoices.

Task:
Classify the customer message as BILLING, TECHNICAL, or ACCOUNT.

Example 1:
"My invoice contains an unexpected charge."
→ BILLING

Example 2:
"The application crashes when I upload a file."
→ TECHNICAL

Example 3:
"I cannot reset my password."
→ ACCOUNT

Now classify:
"I was charged twice for my subscription."

Examples teach behavior, not just formatting

Examples teach behavior, not just formatting works best when treated as a measurable surface. Capture one golden transcript, one failure case, and the rollback note before expanding scope. Record timings and token or query cost next to functional results. Cost visibility early prevents surprise bills when the path moves from demo to shared environments. Budget tokens per turn and per session. Agentic tools expand context aggressively; hard caps keep demos from becoming surprise invoices.

Example selection

Example selection works best when treated as a measurable surface. Capture one golden transcript, one failure case, and the rollback note before expanding scope. Keep configuration outside application code. Environment files, secret stores, and feature flags belong in one place operators can audit without reading the whole graph. Budget tokens per turn and per session. Agentic tools expand context aggressively; hard caps keep demos from becoming surprise invoices. Example selection works best when treated as a measurable surface. Capture one golden transcript, one failure case, and the rollback note before expanding scope. Prefer small, testable units over sprawling scripts. When a step fails, the failure should point at a single responsibility rather than a tangled pipeline.

Example 1: straightforward billing issue
Example 2: straightforward technical issue
Example 3: ambiguous issue
Example 4: boundary case
Example 5: short, poorly written request

Number of examples

For Number of examples, define the inputs, the owner of the step, and the exit criteria before changing code. Operators should be able to re-run the step from a known checkpoint without guessing hidden state. Treat this stage as a contract between inputs and validated outputs. Name the artifacts, define success checks, and refuse silent partial completion. Prefer structured outputs with schema validation over free-form prose when the next step is code or a tool call.

Example diversity

For Example diversity, define the inputs, the owner of the step, and the exit criteria before changing code. Operators should be able to re-run the step from a known checkpoint without guessing hidden state. Record timings and token or query cost next to functional results. Cost visibility early prevents surprise bills when the path moves from demo to shared environments. Prefer structured outputs with schema validation over free-form prose when the next step is code or a tool call.

"can't login"
"I've been locked out since yesterday"
"Password reset isn't sending me anything"
"login broken after changing phone"

Positive and negative examples

For Positive and negative examples, define the inputs, the owner of the step, and the exit criteria before changing code. Operators should be able to re-run the step from a known checkpoint without guessing hidden state. Keep configuration outside application code. Environment files, secret stores, and feature flags belong in one place operators can audit without reading the whole graph. Prefer structured outputs with schema validation over free-form prose when the next step is code or a tool call. For Positive and negative examples, define the inputs, the owner of the step, and the exit criteria before changing code. Operators should be able to re-run the step from a known checkpoint without guessing hidden state. Prefer small, testable units over sprawling scripts. When a step fails, the failure should point at a single responsibility rather than a tangled pipeline.

Input:
"The customer says the package arrived late."
Good output:
"The customer reports that the package arrived late."
Bad output:
"The courier lost the package."
Reason:
The customer did not state that the courier lost it.

Formatting examples

When working through Formatting examples, write down the contract first: required inputs, success signal, and what happens on partial failure. That checklist keeps later code changes honest. Treat this stage as a contract between inputs and validated outputs. Name the artifacts, define success checks, and refuse silent partial completion. Cache stable system instructions and tool schemas. Re-sending identical preamble is a common source of burn.

Issue: ...
Evidence: ...
Action: ...

Input:
"The login token expired after 30 days."
Output:
Issue: Expired authentication token
Evidence: Token expiration is reported after 30 days
Action: Regenerate the authentication token

Consistency between examples

When working through Consistency between examples, write down the contract first: required inputs, success signal, and what happens on partial failure. That checklist keeps later code changes honest. Record timings and token or query cost next to functional results. Cost visibility early prevents surprise bills when the path moves from demo to shared environments. Cache stable system instructions and tool schemas. Re-sending identical preamble is a common source of burn.

Example 1:
"Can't log in" → ACCOUNT
Example 2:
"Can't log in" → TECHNICAL

In-context learning

When working through In-context learning, write down the contract first: required inputs, success signal, and what happens on partial failure. That checklist keeps later code changes honest. Keep configuration outside application code. Environment files, secret stores, and feature flags belong in one place operators can audit without reading the whole graph. Cache stable system instructions and tool schemas. Re-sending identical preamble is a common source of burn. When working through In-context learning, write down the contract first: required inputs, success signal, and what happens on partial failure. That checklist keeps later code changes honest. Prefer small, testable units over sprawling scripts. When a step fails, the failure should point at a single responsibility rather than a tangled pipeline.

Example:
"Password reset email never arrived." → ACCOUNT
User:
"I can't access my account."

Choosing between zero-shot and few-shot

Choosing between zero-shot and few-shot works best when treated as a measurable surface. Capture one golden transcript, one failure case, and the rollback note before expanding scope. Treat this stage as a contract between inputs and validated outputs. Name the artifacts, define success checks, and refuse silent partial completion. Budget tokens per turn and per session. Agentic tools expand context aggressively; hard caps keep demos from becoming surprise invoices.

Breaking a complex task into subtasks

Breaking a complex task into subtasks works best when treated as a measurable surface. Capture one golden transcript, one failure case, and the rollback note before expanding scope. Record timings and token or query cost next to functional results. Cost visibility early prevents surprise bills when the path moves from demo to shared environments. Budget tokens per turn and per session. Agentic tools expand context aggressively; hard caps keep demos from becoming surprise invoices.

Task:
Analyze this customer support case.

Step 1:
Extract the relevant facts.

Step 2:
Identify the customer's actual problem.

Step 3:
List plausible causes supported by the evidence.

Step 4:
Select the safest supported resolution.

Step 5:
Draft the customer-facing response.

Planning before answering

Planning before answering works best when treated as a measurable surface. Capture one golden transcript, one failure case, and the rollback note before expanding scope. Keep configuration outside application code. Environment files, secret stores, and feature flags belong in one place operators can audit without reading the whole graph. Budget tokens per turn and per session. Agentic tools expand context aggressively; hard caps keep demos from becoming surprise invoices. Planning before answering works best when treated as a measurable surface. Capture one golden transcript, one failure case, and the rollback note before expanding scope. Prefer small, testable units over sprawling scripts. When a step fails, the failure should point at a single responsibility rather than a tangled pipeline.

Goal:
Resolve the customer's login problem using only supported evidence.
Plan:
1. Read the latest customer message.
2. Check the relevant conversation history.
3. Identify authentication-related evidence.
4. Compare the evidence with the known troubleshooting guidance.
5. Produce a response that does not claim unsupported facts.

Critique and revision

For Critique and revision, define the inputs, the owner of the step, and the exit criteria before changing code. Operators should be able to re-run the step from a known checkpoint without guessing hidden state. Treat this stage as a contract between inputs and validated outputs. Name the artifacts, define success checks, and refuse silent partial completion. Prefer structured outputs with schema validation over free-form prose when the next step is code or a tool call.

First:
Draft the response.

Then:
Check the draft for:
- unsupported claims,
- missing evidence,
- incorrect troubleshooting steps,
- promises the support team cannot make,
- failure to answer the customer's actual question.

Finally:
Revise the response to correct any problems.

Reasoning-oriented prompting

For Reasoning-oriented prompting, define the inputs, the owner of the step, and the exit criteria before changing code. Operators should be able to re-run the step from a known checkpoint without guessing hidden state. Record timings and token or query cost next to functional results. Cost visibility early prevents surprise bills when the path moves from demo to shared environments. Prefer structured outputs with schema validation over free-form prose when the next step is code or a tool call.

Identify the relevant facts.
Determine which facts support each possible conclusion.
Eliminate conclusions that require unsupported assumptions.
Select the conclusion supported by the evidence.

classification: password_reset_issue
evidence_supported: true
recommended_action: resend_reset_link
confidence: high

Self-consistency: don’t trust a single path blindly

For Self-consistency: don’t trust a single path blindly, define the inputs, the owner of the step, and the exit criteria before changing code. Operators should be able to re-run the step from a known checkpoint without guessing hidden state. Keep configuration outside application code. Environment files, secret stores, and feature flags belong in one place operators can audit without reading the whole graph. Prefer structured outputs with schema validation over free-form prose when the next step is code or a tool call. For Self-consistency: don’t trust a single path blindly, define the inputs, the owner of the step, and the exit criteria before changing code. Operators should be able to re-run the step from a known checkpoint without guessing hidden state. Prefer small, testable units over sprawling scripts. When a step fails, the failure should point at a single responsibility rather than a tangled pipeline.

When prompting should stop and computation should begin

When working through When prompting should stop and computation should begin, write down the contract first: required inputs, success signal, and what happens on partial failure. That checklist keeps later code changes honest. Treat this stage as a contract between inputs and validated outputs. Name the artifacts, define success checks, and refuse silent partial completion. Cache stable system instructions and tool schemas. Re-sending identical preamble is a common source of burn.

Operational checklist

When working through Operational checklist, write down the contract first: required inputs, success signal, and what happens on partial failure. That checklist keeps later code changes honest.

Document the happy path and the recovery path together. Retries, human gates, and dead-letter handling are part of the product, not later polish.

Cache stable system instructions and tool schemas. Re-sending identical preamble is a common source of burn.

Keep render work cheap and push expensive derivation behind memoization only after measuring. Premature memo can hide stale props bugs.

Add a smoke test that exercises the critical path in CI with fixtures, not live paid APIs, whenever budgets allow.

Record timings and token or query cost next to functional results. Cost visibility early prevents surprise bills when the path moves from demo to shared environments.

Before promoting the stack, freeze versions, capture a golden transcript for the critical path, and confirm rollback steps. Shared environments need rate limits, tenancy checks, and a clear owner for secret rotation. Prefer boring reliability over clever one-off demos.

Batch note for 96fc8ed96065: keep provider keys out of the repo, set a per-session token ceiling, and store transcripts next to the eval fixtures so later model swaps stay comparable.