Task Classifier
SkillProductivityClassifies a user request into a complexity tier (T1 simple, T2 standard, T3 complex) to feed the model-router. Exposes a numeric score 0-100 and an explanation. Triggers when the user asks "classifie cette demande", "evalue la complexite", "quel est le niveau de cette t
Use Task Classifier in Claude, ChatGPT or Ahel Desktop
Free. Sign in, add Task Classifier and connect your AI. About a minute.
Also: Claude Code · Cursor · Codex
Then ask your AI: use the Task Classifier skill
Details
Instructions available. Your AI can read the instructions. Execution depends on the setup they require.
Account requirements not reviewed. Check the skill instructions before use; Ahel provides instructions and does not run this skill.
No other account needed.
Add Ahel to your AI once: Claude, ChatGPT, Cursor, Claude Code or Codex. Then ask it to use this.
What this skill tells your AI
The instructions your AI receives, as published by geekfamilycorp/tricorderkit in plugins/token-optimizer/skills/task-classifier/SKILL.md and read by Ahel’s review.
Analyse la complexite cognitive d'une requete et retourne un tier avec justification.
Entrees
Le prompt de l'utilisateur, eventuellement enrichi du contexte precedent.
Dimensions evaluees
-
Longueur input (0-30 pts)
- < 500 tokens : 0
- 500-1500 : 5
- 1500-5000 : 10
- 5000-15000 : 20
-
15000 : 30
-
Longueur output attendue (0-20 pts)
- Reponse courte (< 300 mots) : 0
- Reponse moyenne (300-1500 mots) : 10
- Reponse longue (> 1500 mots ou code > 200 lignes) : 20
-
Profondeur de raisonnement (0-25 pts)
- Factuel / extraction : 0
- Analyse simple : 10
- Raisonnement multi-etapes : 20
- Architecture / trade-offs / preuve : 25
-
Domaine sensible (0-15 pts)
- Aucun : 0
- Technique standard : 5
- Legal / medical / securite / financier / production : 15
-
Contexte multi-sources (0-10 pts)
- Mono-source : 0
- 2-3 sources : 5
-
3 sources ou multi-fichier : 10
Score final → Tier
- 0-25 pts : T1 Haiku
- 26-60 pts : T2 Sonnet
- 61-100 pts : T3 Opus
Usage programmatique
Pour une classification deterministe, appeler :
python3 ${CLAUDE_PLUGIN_ROOT}/scripts/classify.py --prompt "<prompt>" --json
Sortie :
{
"score": 42,
"tier": "T2",
"model": "claude-sonnet-4-6",
"dimensions": {
"input_length": 10,
"output_length": 10,
"reasoning_depth": 20,
"sensitive_domain": 0,
"multi_source": 2
},
"explanation": "Tache de redaction moyenne avec analyse..."
}
Sortie a restituer a l'utilisateur
Format court :
Score : 42/100 -> T2 Sonnet 4.6
Raison : redaction moyenne avec analyse documentaire (input 3k, output ~1k, analyse multi-etapes).
Signals
- GitHub stars
- 30
- Forks
- 1
- Last commit
- Aug 2026
Advanced
- Item type
- skill
- Key
task-classifier- Source
- github.com/geekfamilycorp/tricorderkit
github.com/geekfamilycorp/tricorderkit
More in Productivity
Skill · coreyhaines31
More in Productivitygws-calendar
Skill · googleworkspace
More in Productivitylark-minutes
Skill · larksuite
More in Productivitylark-workflow-standup-report
Skill · larksuite
More in Productivityenergy-procurement
Skill · affaan-m
More in Productivityhomelab-pihole-dns
Skill · affaan-m
More in Productivity