Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx skills add khalilbenaz/claude-skills-collection --skill supervisor-buildergit clone --depth 1 https://github.com/khalilbenaz/claude-skills-collectionWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/khalilbenaz/claude-skills-collection/supervisor-builder)<a href="https://agentmods.dev/skills/khalilbenaz/claude-skills-collection/supervisor-builder"><img src="https://agentmods.dev/badge/skills/khalilbenaz/claude-skills-collection/supervisor-builder/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/khalilbenaz/claude-skills-collection/supervisor-builder"><img src="https://agentmods.dev/badge/skills/khalilbenaz/claude-skills-collection/supervisor-builder.svg" alt="Reviewed on agentmods" width="80" height="20"></a>- NVIDIA SkillSpector pass
What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00080 | $0.03647 |
| Opus 5.5 | $0.00032 | $0.01459 |
| Sonnet 5.5 | $0.00016 | $0.00729 |
| Haiku 4.5 | $0.00008 | $0.00365 |
Grade A, and why
supervisor-builder scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 2mo ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 362 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Agent Supervisor Builder
Quand utiliser ce skill
Utiliser ce skill quand une architecture multi-agents requiert une couche de contrôle active : garantie qualité dynamique, interruption d'un agent défaillant, redistribution de charge, escalade humaine. Indispensable dès que deux sous-agents ou plus s'exécutent en parallèle ou en chaîne avec des SLAs à respecter.
Critères de décision — quel pattern choisir ?
| Besoin | Pattern recommandé |
|---|---|
| Délégation simple, un agent à la fois | create_supervisor LangGraph (step 9) |
| Équipe hiérarchique avec rôles fixes | CrewAI Process.hierarchical |
| Intervenir en cours d'exécution | Monitor + CorrectionLoop (steps 3-5) |
| Failover automatique + retry | InterventionRules + LoadBalancer (steps 4, 7) |
| Escalade humaine obligatoire | EscalationManager (step 6) |
Workflow
1. Définir les actions du supervisor
Commencer par énumérer toutes les actions possibles avant d'écrire une ligne de logique :
from enum import Enum
class SupervisorAction(Enum):
DISPATCH = "dispatch" # Assigner une tâche à un sous-agent
MONITOR = "monitor" # Surveiller la progression
INTERVENE = "intervene" # Envoyer une correction en cours d'exécution
REDIRECT = "redirect" # Réassigner vers un autre sous-agent
TERMINATE = "terminate" # Arrêter un agent défaillant
ESCALATE = "escalate" # Remonter à un humain ou orchestrateur supérieur
APPROVE = "approve" # Valider un output avant livraison
2. Routing intelligent — choisir le bon sous-agent
Classifier l'intention, scorer les capacités, fallback vers le plus généraliste.
class IntelligentRouter:
def __init__(self, agents: list):
self.agents = {a.id: a for a in agents}
def route(self, request: str) -> str:
intent = self._classify_intent(request)
best = max(
self.agents.values(),
key=lambda a: self._score(a, intent)
)
return best.id
def _classify_intent(self, request: str) -> dict:
# Appel LLM léger (gpt-4o-mini suffit pour la classification)
prompt = f"Classifie: '{request}'. JSON: task_type, domain, complexity(low|med|high)"
return llm.invoke(prompt, response_format="json")
def _score(self, agent, intent: dict) -> float:
return (
1.0 * (intent["task_type"] in agent.capabilities) +
0.5 * (intent["domain"] in agent.domains) +
0.3 * (agent.complexity_level >= intent["complexity"])
)
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 2mo ago First seen · 362 lines · 80 tokens per session scan A faa8afc4d04f
supervisor-builder is a skill published in the GitHub repository khalilbenaz/claude-skills-collection (27 stars, last pushed 2mo ago), licensed MIT. It adds 80 tokens to every session and 3,647 once invoked, about $0.0003 per session on Opus 5.5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
supervisor-builder compared with skills from other repositories
| Mod | Type | Stars | Last push | Licence | Scan | Tokens | Works with |
|---|---|---|---|---|---|---|---|
| supervisor-builderkhalilbenaz/claude-skills-collection | Skill | 27 | 2mo ago | MIT | A | 80 | Claude Code, Codex |
| explainsoftspark/ai-toolkit | Skill | 179 | 10d ago | Apache-2.0 | A | 34 | Claude Code |
| finance-econ-literacymodu-ai/moai-cowork | Skill | 315 | 29d ago | Apache-2.0 | A | 128 | Claude Code |
| common-sense-index-investing-boglesimbajigege/book2skills | Skill | 203 | 2mo ago | MIT | A | 38 | Claude Code, Codex |
| stock-analysis-leadjohnqtcg/awesome-skills | Skill | 30 | 29d ago | MIT | A | 229 | Claude Code |
| stock-business-reviewjohnqtcg/awesome-skills | Skill | 30 | 29d ago | MIT | A | 121 | Claude Code |
| stock-earnings-quality-reviewjohnqtcg/awesome-skills | Skill | 30 | 29d ago | MIT | A | 147 | Claude Code |