Exemples catégoriels
Montrez des exemples de ce qu’il faut signaler et de ce qu’il faut ignorer.
Exemples catégoriels est une leçon Claude Architect gratuite sur CoddyKit. Ceci est la leçon 2 sur 4. Tu peux lire la leçon complète ci-dessous gratuitement — puis la pratiquer en direct dans le navigateur avec un éditeur de code intégré et un tuteur IA 24/7. Elle fait partie du parcours d'apprentissage Claude Architect, et ta progression se synchronise sur le web et l'application CoddyKit. Le cours Claude Architect comprend 4 leçons au total.
Certaines parties de cette leçon n'ont pas encore été traduites et s'affichent en anglais.
Why Categorical Examples Matter
When you give Claude a job that has a fuzzy boundary, vague instructions like "be more precise" rarely work. The model can't read your mind about where the line sits.
The reliable fix is categorical examples: a small set of 2-4 few-shot examples that show concrete cases of what to report and what to ignore. The model generalizes from these examples to new, unseen inputs — it does not merely repeat them.
This lesson shows how to build those examples for a classic architect task: deciding what is worth flagging and what is noise.
Explicit Criteria Beat Vague Adjectives
Before you even reach for examples, state an explicit criterion. Explicit rules consistently beat vague adjectives.
- Vague: "Review the code carefully."
- Explicit: "Flag a comment only when it contradicts the code it describes."
The explicit version draws a sharp line. Categorical examples then make that line unmistakable by pinning it to concrete cases on each side.
system = (
"You review code comments. "
"Flag a comment ONLY when it contradicts the code it describes. "
"Ignore style, tone, and outdated-but-harmless notes."
)The Shape of a Categorical Example Set
A good set has examples on both sides of the boundary, each with a one-line reason. Two REPORT cases and two IGNORE cases is enough for most ambiguities — 2-4 targeted examples is the sweet spot.
Each example carries three parts:
- the input (the thing being judged)
- the verdict (report / ignore)
- a short reason tied to your criterion
The reason is what lets the model generalize correctly instead of pattern-matching on surface features.
Worked Example: Report vs Ignore
Here is a categorical example block for the comment-review task. Notice the symmetry: contradictions are reported; harmless mismatches are ignored.
EXAMPLES = """
<example verdict="REPORT">
code: return price * 1.2
comment: # applies 10% tax
reason: comment says 10% but code applies 20% (contradiction)
</example>
<example verdict="IGNORE">
code: timeout = 30 # seconds
comment: # seconds
reason: accurate; matches the code
</example>
<example verdict="IGNORE">
code: # TODO: refactor later
comment: # TODO: refactor later
reason: stylistic note, not a contradiction
</example>
<example verdict="REPORT">
code: if user.is_admin: deny()
comment: # allow admins
reason: comment says allow but code denies (contradiction)
</example>
"""Generalization, Not Memorization
The power of categorical examples is that the model generalizes. It does not need an example for every possible input.
From the four cases above, Claude learns the rule "flag semantic contradictions, ignore harmless mismatches" and applies it to a brand-new comment it has never seen, such as a docstring that claims a function returns a list when it returns a dict.
This is why few-shot examples are ideal for consistency, edge cases, output format, and reducing hallucination.
Pin the Edges of the Boundary
The most valuable examples sit right at the boundary, where the decision is genuinely hard. A near-miss IGNORE next to a near-hit REPORT teaches far more than two obvious cases.
For a research agent deciding which statistics to surface, you might contrast a stale figure with a current one — and note that dates often resolve apparent contradictions rather than one number simply being wrong.
EXAMPLES = """
<example verdict="REPORT">
fact: "Revenue was $4.1B in FY2024" (source dated 2025-02)
reason: current, sourced, on-topic -> surface it
</example>
<example verdict="IGNORE">
fact: "Revenue was $3.2B" (source dated 2019, no fiscal year)
reason: stale and undated for our scope -> drop, do not treat as a conflict
</example>
"""Examples Drive Structured Output Too
Categorical examples pair naturally with structured output. When the verdict must be machine-readable, force a tool call so the result is valid JSON every time.
Use tool_choice of type "any" to guarantee the model calls some tool (guaranteeing structured output), or force one specific tool by name. A JSON Schema then eliminates syntax errors and enforces required fields.
tools = [{
"name": "record_finding",
"description": "Record a review verdict for one item.",
"input_schema": {
"type": "object",
"properties": {
"verdict": {"enum": ["report", "ignore"]},
"reason": {"type": "string"}
},
"required": ["verdict", "reason"]
}
}]
resp = client.messages.create(
model="claude-sonnet-4-5",
max_tokens=512,
tools=tools,
tool_choice={"type": "any"},
system=system,
messages=[{"role": "user", "content": EXAMPLES + new_item}],
)Only Require Fields That Always Exist
A subtle trap: in your finding schema, mark a field required only if it is always present. If you require a field that may be absent — say a line_number for findings that aren't line-specific — the model will fabricate a value to satisfy the schema.
For categories that may grow, use an enum plus an "other" value and a free-text detail field. That keeps the output structured while staying extensible.
"category": {"enum": ["contradiction", "security", "other"]},
"category_detail": {"type": "string"} # filled when category == other
# line_number is NOT required: many findings are file-levelWhen Examples Are the Wrong Tool
Categorical examples sharpen a judgment boundary. They do not invent missing information.
If the model returns an empty or wrong result because the needed fact is simply absent from the source, more examples won't help — and neither will retry-with-feedback. Retry fixes format, structural, and arithmetic errors, not absent data.
Distinguish a genuine empty result ("no matches found") from an access failure (the source couldn't be read). Treat them differently in your recovery logic.
Categorical Examples in CLAUDE.md and Reviews
Bake your report/ignore policy where the work happens. For a CI review job, put the examples in project-level CLAUDE.md (shared via VCS) so every run and every teammate inherits the same boundary.
In CI, run the review in an isolated session (less biased than the generation context) with -p for non-interactive output. Explicit criteria plus categorical examples are exactly how you minimize false positives.
## Review policy (categorical)
REPORT: a comment that contradicts the code it describes.
IGNORE: style nits, tone, outdated-but-harmless TODOs.
Example REPORT -> `# 10% tax` over `price * 1.2`
Example IGNORE -> `timeout = 30 # seconds` (accurate)Putting It Together
The full recipe for a clean report/ignore decision:
- Write one explicit criterion for the boundary.
- Add 2-4 categorical examples split across both sides, each with a reason.
- Pin examples at the hard edges, not the obvious cases.
- Force structured output with a tool; require only ever-present fields.
- Remember: examples shape judgment, not missing data.
Do this and Claude generalizes your intent reliably across inputs it has never seen.
Quick Check
A review agent flags too many harmless comments as problems. You want it to report only genuine contradictions. Which change will most reliably tighten the boundary?
Recap
Key takeaways:
- Explicit criteria + categorical examples beat vague adjectives for fuzzy boundaries.
- Use 2-4 examples split across report and ignore, each with a reason; the model generalizes, it doesn't just repeat.
- Pin examples at the hard edges; remember dates often resolve apparent contradictions.
- Pair with structured output (
tool_choice"any" + JSON Schema); require only ever-present fields, use enum+"other" for extensibility. - Examples sharpen judgment, not missing data — absent facts need a different fix, and retry only repairs format/arithmetic errors.
Apprends Python avec un tuteur IA — gratuit
Écris et exécute du vrai code dans ton navigateur, obtiens de l'aide instantanée d'un tuteur IA disponible 24h/24, et reprends là où tu t'es arrêté sur le web ou dans l'app.
- Cours
- 26
- Leçons
- 104
Questions Fréquemment Posées
La leçon « Exemples catégoriels » est-elle gratuite ?
Oui — le texte complet de « Exemples catégoriels » est gratuit à lire ici sur le web. Pour la pratiquer de manière interactive (un éditeur de code intégré et un tuteur IA 24/7) et déverrouiller le reste du cours Claude Architect, passe à CoddyKit PRO. Le cours Claude Architect comprend 4 leçons au total.
Qu'est-ce que j'apprendrai dans « Exemples catégoriels » ?
Montrez des exemples de ce qu’il faut signaler et de ce qu’il faut ignorer. Tu pratiques Claude Architect avec du code pratique que tu exécutes directement dans le navigateur, et un tuteur IA 24/7 répond à tes questions au fur et à mesure que tu avances dans la leçon.
Dois-je avoir de l'expérience pour commencer Claude Architect ?
Aucune expérience préalable n'est requise. Claude Architect sur CoddyKit est structuré pour les débutants jusqu'aux apprenants avancés, donc tu peux commencer ici ou depuis le début et avancer à ton rythme. Ceci est la leçon 2 sur 4.
Combien de temps prend la leçon « Exemples catégoriels » ?
La plupart des leçons CoddyKit prennent environ 5–10 minutes. Chacune est courte et interactive, tu progresses régulièrement et tu repiques exactement où tu t'es arrêté sur le web et l'app.
Peux-tu écrire et exécuter du code dans cette leçon Claude Architect ?
Oui. Chaque leçon Claude Architect inclut un éditeur de code intégré, tu écris et exécutes du vrai code directement dans ton navigateur et tu reçois des retours IA instantanés — aucune configuration locale requise.
Toutes les leçons de ce cours
- Critères explicites plutôt qu’instructions vagues
- Exemples catégoriels
- Critères de gravité avec exemples
- Réduire les faux positifs