📦

learn

Name: learn
Author: agentskill-sh

v2.0 Révision du contenu r1 Critique 🌐 Accès réseau📁 Accès au système de fichiers⚡ Contient des scripts⚙️ Commandes externes🔑 Variables d’environnement

Installer des compétences d'agent en cours de conversation

Les agents ont souvent besoin d'instructions spécialisées qui ne sont pas chargées au début d'une session. Cette compétence permet à Claude, Codex et Claude Code de rechercher, prévisualiser, analyser, installer, mettre à jour, supprimer et évaluer des compétences depuis agentskill.sh.

Prend en charge: Claude Codex Code(CC)

⚠️ 38 Médiocre

Maintain this Skill? Claim it ⚠️ 38 Médiocre

N’installez pas automatiquement ce Skill.

La politique de référence exige un examen par un opérateur avant toute action d’installation.

Installer avec mon Agent

Copiez cette demande dans votre Agent. Elle contient la page Skill de référence et le manifeste.

Demande d’agent

Review the Skillstore skill "learn" from https://skillstore.io/skills/agentskill-sh-learn.md and its manifest at https://skillstore.io/api/skills/agentskill-sh-learn/manifest. Verify the artifact. Do not auto-install. Inspect the skill and report your findings, then wait for an operator or manual installation decision.

Votre Agent doit toujours présenter son plan et demander toute confirmation exigée par la politique de sécurité.

Ressources lisibles par les agents

Utilisez ces liens lorsqu’un Agent IA, un crawler ou un script a besoin d’un contexte propre au lieu de lire toute la page.

Détail Markdown GET /skills/agentskill-sh-learn.md manifest signé GET /api/skills/agentskill-sh-learn/manifest lockfile signé GET /api/skills/agentskill-sh-learn/lockfile

Tester

Utilisation de « learn ». /learn documentation

Résultat attendu:

Un tableau classé des compétences correspondantes avec les auteurs, les nombres d'installations, les scores de sécurité, de courtes descriptions et un choix d'installation interactif.

Utilisation de « learn ». /learn scan ./SKILL.md

Résultat attendu:

Un rapport de sécurité qui liste le score, les problèmes détectés, les points de terminaison réseau, l'accès aux fichiers, les commandes shell et une recommandation d'autorisation ou de blocage.

Utilisation de « learn ». /learn update

Résultat attendu:

Un résumé des compétences installées, de l'état des mises à jour, des résultats de la nouvelle analyse de sécurité et une demande de confirmation avant l'application des mises à jour.

Audit de sécurité

Critique

v7 • 09/07/2026 Ouvrir le rapport versionné

Most static hits are false positives from documentation and defensive scanner examples, especially references/SECURITY.md and Markdown inline-code in SKILL.md. The real risks are the intended remote marketplace behavior: fetching and installing remote skill content into agent directories, self-updating the installer, and sending feedback or scan reports to agentskill.sh.

Fichiers analysés

1,630

Lignes analysées

115

Éléments à examiner

Faux positifs ignorés

Préoccupations de sécurité confirmées (8)

Critique

multiple:1

[HEURISTIC] DANGEROUS COMBINATION: Code execution + Network + Credential access

This combination is common in credential stealers and RATs

The aggregate combination is real because the skill combines external marketplace fetching with writes to agent skill directories and optional telemetry. The intent is legitimate, but the combined behavior creates supply-chain risk.

Critique

multiple:1

[HEURISTIC] DANGEROUS COMBINATION: Network + Credentials + Evasion techniques

This combination suggests data exfiltration with anti-detection

Élevé

multiple:1

[HEURISTIC] SUSPICIOUS COMBINATION: Code execution + Persistence mechanism

This combination is common in malware that maintains access

Élevé

multiple:1

[HEURISTIC] SUSPICIOUS COMBINATION: Filesystem + Credentials + Network

This combination could indicate credential harvesting and exfiltration

Élevé

SKILL.md:256-331

Remote Skill Installation Can Change Agent Behavior

The install flow fetches skill content from agentskill.sh, scans it, and writes it into the detected agent skill directory. A compromised marketplace response or weak scan result could install instructions that influence future agent behavior.

The cited install flow explicitly fetches remote skill content, scans it, writes a skill file, and tracks the install. This is intended behavior, but it is a clear supply-chain control point.

Élevé

SKILL.md:560-568

Self-Update Threshold Is Weaker Than Install Block Threshold

The self-update flow says to overwrite the current learn skill if the new version scan passes with score at least 50, while the install flow blocks skills below 70. This could permit a lower-scoring update to replace the installer itself.

The lines directly describe fetching the latest version and overwriting the current skill when scan score is at least 50. The same file documents a stricter install block threshold elsewhere.

Moyen

SKILL.md:772

System reconnaissance

| Invalid score | "Score must be an integer between 1 and 5."

This section describes runtime installation, update, or feedback behavior that fetches remote content, writes skill files, or sends rating data. The behavior is purposeful, but it has supply-chain and privacy impact.

Moyen

SKILL.md:634-760

Opt-Out Rating and Feedback Telemetry

The skill instructs agents to rate installed skills by default, send ratings after a short confirmation window, and avoid ending a session with unrated skills. This sends platform and agent-generated comments to an external service unless disabled.

The cited section describes automatic rating, a 5-second default-send path, platform data, comments, and session-end rating behavior. It is disclosed, but it is still privacy-relevant external telemetry.

Éléments d’évaluation des capacités (115)

Il s’agit de capacités locales réelles qui peuvent être attendues pour cette compétence ; elles nécessitent donc un examen, mais ne sont pas comptabilisées comme comportement malveillant confirmé.

Élevé

SKILL.md:582

Hidden file in home directory

1. Check if `.openclaw/` directory exists OR `~/.openclaw/workspace/` exists → **OpenClaw**

The skill detects or writes agent skill directories, which can alter future agent behavior. This is intended functionality, but writing remote skill content into hidden agent configuration paths is a real security-relevant filesystem action.

Élevé

SKILL.md:583

Hidden file in home directory

- Install path: `~/.openclaw/workspace/skills/<slug>.md`

Moyen

SKILL.md:40-44

Ruby/shell backtick execution

1. Use WebFetch to call: `https://agentskill.sh/api/agent/search?q=<URL-encoded query>&limit=5`

SKILL.md instructs the agent to contact agentskill.sh APIs for search, install, update, tracking, reports, or feedback. The domain matches the stated marketplace purpose, but it is real external network behavior.

Moyen

SKILL.md:77-81

Ruby/shell backtick execution

2. Use WebFetch to call: `https://agentskill.sh/api/agent/skills/<slug>/install?owner=<owner>`

Moyen

SKILL.md:87-88

Ruby/shell backtick execution

1. Parse the slug from the URL path (last segment of `https://agentskill.sh/<slug>`)

Moyen

SKILL.md:88-91

Ruby/shell backtick execution

2. Use WebFetch to call: `https://agentskill.sh/api/agent/skills/<slug>/install`

Moyen

SKILL.md:121-125

Ruby/shell backtick execution

1. Use WebFetch to call: `https://agentskill.sh/api/agent/search?section=trending&limit=5`

Moyen

SKILL.md:133-134

Ruby/shell backtick execution

3. Use WebFetch to POST to `https://agentskill.sh/api/skills/<slug>/agent-feedback` with JSON body:

Moyen

SKILL.md:179-180

Ruby/shell backtick execution

2. Collect all slugs and call the batch version endpoint: `https://agentskill.sh/api/agent/skills/ve

Moyen

SKILL.md:256-270

Ruby/shell backtick execution

1. Fetch skill content from `https://agentskill.sh/api/agent/skills/<slug>/install?platform=<platfor

Moyen

SKILL.md:270-280

Ruby/shell backtick execution

```

Moyen

SKILL.md:280-284

Ruby/shell backtick execution

```

Moyen

SKILL.md:284-296

Ruby/shell backtick execution

```

Moyen

SKILL.md:296-318

Ruby/shell backtick execution

```

Moyen

SKILL.md:318-329

Ruby/shell backtick execution

```

Moyen

SKILL.md:329-331

Ruby/shell backtick execution

```

Moyen

SKILL.md:331-333

Ruby/shell backtick execution

9. Track the install — use WebFetch to POST to `https://agentskill.sh/api/skills/<slug>/install` wit

Moyen

SKILL.md:333-338

Ruby/shell backtick execution

```json

Moyen

SKILL.md:338-344

Ruby/shell backtick execution

```

Moyen

SKILL.md:344-347

Ruby/shell backtick execution

```

Moyen

SKILL.md:347-354

Ruby/shell backtick execution

**Location:** `<install path>`

Moyen

SKILL.md:354-355

Ruby/shell backtick execution

Rate this skill later: `/learn feedback <slug> <1-5> [optional comment]`

Moyen

SKILL.md:355-371

Ruby/shell backtick execution

```

Moyen

SKILL.md:556-560

Ruby/shell backtick execution

Before executing any subcommand, check if this `/learn` skill itself is up to date.

Moyen

SKILL.md:560

Ruby/shell backtick execution

1. Read the current `/learn` skill file and extract the `contentSha` from the metadata header

Moyen

SKILL.md:561-562

Ruby/shell backtick execution

2. Use WebFetch to call: `https://agentskill.sh/api/agent/skills/learn/version`

Moyen

SKILL.md:562

Ruby/shell backtick execution

3. Compare the local `contentSha` with the remote `contentSha`

Moyen

SKILL.md:565-582

Ruby/shell backtick execution

a. Fetch the latest version from `https://agentskill.sh/api/agent/skills/learn/install`

Moyen

SKILL.md:582

Ruby/shell backtick execution

1. Check if `.openclaw/` directory exists OR `~/.openclaw/workspace/` exists → **OpenClaw**

Moyen

SKILL.md:583-584

Ruby/shell backtick execution

- Install path: `~/.openclaw/workspace/skills/<slug>.md`

Moyen

SKILL.md:584-585

Ruby/shell backtick execution

2. Check if `.claude/` directory exists in the project root → **Claude Code / Claude Desktop**

Moyen

SKILL.md:585-586

Ruby/shell backtick execution

- Install path: `.claude/skills/<slug>.md`

Moyen

SKILL.md:586-587

Ruby/shell backtick execution

3. Check if `.cursor/` directory exists → **Cursor**

Moyen

SKILL.md:587-588

Ruby/shell backtick execution

- Install path: `.cursor/skills/<slug>.md`

Moyen

SKILL.md:588-589

Ruby/shell backtick execution

4. Check if `.github/copilot/` directory exists → **GitHub Copilot**

Moyen

SKILL.md:589-590

Ruby/shell backtick execution

- Install path: `.github/copilot/skills/<slug>.md`

Moyen

SKILL.md:590-591

Ruby/shell backtick execution

5. Check if `.windsurf/` directory exists → **Windsurf**

Moyen

SKILL.md:591-592

Ruby/shell backtick execution

- Install path: `.windsurf/skills/<slug>.md`

Moyen

SKILL.md:592-593

Ruby/shell backtick execution

6. Check if `.cline/` directory exists → **Cline**

Moyen

SKILL.md:593-594

Ruby/shell backtick execution

- Install path: `.cline/skills/<slug>.md`

Moyen

SKILL.md:594-595

Ruby/shell backtick execution

7. Check if `.codex/` directory exists → **Codex**

Moyen

SKILL.md:595-596

Ruby/shell backtick execution

- Install path: `.codex/skills/<slug>.md`

Moyen

SKILL.md:596-597

Ruby/shell backtick execution

8. Check if `.opencode/` directory exists → **OpenCode**

Moyen

SKILL.md:597-598

Ruby/shell backtick execution

- Install path: `.opencode/skills/<slug>.md`

Moyen

SKILL.md:598-599

Ruby/shell backtick execution

9. Check if `.aider/` directory exists → **Aider**

Moyen

SKILL.md:599-600

Ruby/shell backtick execution

- Install path: `.aider/skills/<slug>.md`

Moyen

SKILL.md:600-601

Ruby/shell backtick execution

10. Check if `.gemini/` directory exists → **Gemini CLI**

Moyen

SKILL.md:601-602

Ruby/shell backtick execution

- Install path: `.gemini/skills/<slug>.md`

Moyen

SKILL.md:602-603

Ruby/shell backtick execution

11. Check if `.amp/` directory exists → **Amp**

Moyen

SKILL.md:603-604

Ruby/shell backtick execution

- Install path: `.amp/skills/<slug>.md`

Moyen

SKILL.md:604-605

Ruby/shell backtick execution

12. Check if `.goose/` directory exists → **Goose**

Moyen

SKILL.md:605-606

Ruby/shell backtick execution

- Install path: `.goose/skills/<slug>.md`

Moyen

SKILL.md:606-607

Ruby/shell backtick execution

13. Check if `.roo-code/` directory exists → **Roo Code**

Moyen

SKILL.md:607-608

Ruby/shell backtick execution

- Install path: `.roo-code/skills/<slug>.md`

Moyen

SKILL.md:608-609

Ruby/shell backtick execution

14. Check if `.trae/` directory exists → **Trae**

Moyen

SKILL.md:609-615

Ruby/shell backtick execution

- Install path: `.trae/skills/<slug>.md`

Moyen

SKILL.md:615

Ruby/shell backtick execution

| `.openclaw/` | `openclaw` |

Moyen

SKILL.md:616

Ruby/shell backtick execution

| `.claude/` | `claude-code` |

Moyen

SKILL.md:617

Ruby/shell backtick execution

| `.cursor/` | `cursor` |

Moyen

SKILL.md:618

Ruby/shell backtick execution

| `.github/copilot/` | `copilot` |

Moyen

SKILL.md:619

Ruby/shell backtick execution

| `.windsurf/` | `windsurf` |

Moyen

SKILL.md:620

Ruby/shell backtick execution

| `.cline/` | `cline` |

Moyen

SKILL.md:621

Ruby/shell backtick execution

| `.codex/` | `codex` |

Moyen

SKILL.md:622

Ruby/shell backtick execution

| `.opencode/` | `opencode` |

Moyen

SKILL.md:623

Ruby/shell backtick execution

| `.aider/` | `aider` |

Moyen

SKILL.md:624

Ruby/shell backtick execution

| `.gemini/` | `gemini-cli` |

Moyen

SKILL.md:625

Ruby/shell backtick execution

| `.amp/` | `amp` |

Moyen

SKILL.md:626

Ruby/shell backtick execution

| `.goose/` | `goose` |

Moyen

SKILL.md:627

Ruby/shell backtick execution

| `.roo-code/` | `roo-code` |

Moyen

SKILL.md:628

Ruby/shell backtick execution

| `.trae/` | `trae` |

Moyen

SKILL.md:630-660

Ruby/shell backtick execution

When creating the skill directory, create it if it doesn't exist (e.g., `mkdir -p .claude/skills/`).

Moyen

SKILL.md:660-663

Ruby/shell backtick execution

```

Moyen

SKILL.md:663-664

Ruby/shell backtick execution

Send rating? (Y/n) — disable auto-rating: `/learn config autorating off`

Moyen

SKILL.md:664-666

Ruby/shell backtick execution

```

Moyen

SKILL.md:666-674

Ruby/shell backtick execution

```json

Moyen

SKILL.md:674-675

Ruby/shell backtick execution

```

Moyen

SKILL.md:675-686

Ruby/shell backtick execution

4. **User can override anytime**: `/learn feedback <slug> <score> [comment]`

Moyen

SKILL.md:686-688

Ruby/shell backtick execution

```

Moyen

SKILL.md:688-690

Ruby/shell backtick execution

```

Moyen

SKILL.md:690-706

Ruby/shell backtick execution

Re-enable with: `/learn config autorating on`

Moyen

SKILL.md:706-751

Ruby/shell backtick execution

- `/learn feedback <slug> <new-score> [comment]`

Moyen

SKILL.md:751-757

Ruby/shell backtick execution

```

Moyen

SKILL.md:757-758

Ruby/shell backtick execution

*Override any rating: `/learn feedback <slug> <score> [comment]`*

Moyen

SKILL.md:758-781

Ruby/shell backtick execution

```

Moyen

SKILL.md:781-785

Ruby/shell backtick execution

All endpoints are on `https://agentskill.sh`.

Moyen

SKILL.md:785-786

Ruby/shell backtick execution

| `/api/agent/search?q=<query>&limit=5` | GET | Search skills

Moyen

SKILL.md:786-787

Ruby/shell backtick execution

| `/api/agent/skills/<slug>/install` | GET | Get skill content for installation

Moyen

SKILL.md:787-788

Ruby/shell backtick execution

| `/api/agent/skills/<slug>/version` | GET | Get content SHA for version check

Moyen

SKILL.md:788-789

Ruby/shell backtick execution

| `/api/agent/skills/version?slugs=<csv>` | GET | Batch version check

Moyen

SKILL.md:789-790

Ruby/shell backtick execution

| `/api/skills/<slug>/install` | POST | Track install event

Moyen

SKILL.md:790

Ruby/shell backtick execution

| `/api/skills/<slug>/agent-feedback` | POST | Submit score and comment (include `autoRated: t

Moyen

SKILL.md:582

Hidden file access

1. Check if `.openclaw/` directory exists OR `~/.openclaw/workspace/` exists → **OpenClaw**

Moyen

SKILL.md:583

Hidden file access

- Install path: `~/.openclaw/workspace/skills/<slug>.md`

Faible

SKILL.md:451

Python HTTP libraries

- Manual: check for `os.system`, `subprocess(shell=True)`, `exec`, `eval`, `pickle.loads`, `requests

Faible

SKILL.md:771

Python HTTP libraries

| Rate limited (429) | "Too many requests. Please wait a moment and try again."

Faible

SKILL.md:11

Hardcoded URL

Discover, install, and manage AI agent skills from [agentskill.sh](https://agentskill.sh). This skil

Faible

SKILL.md:40

Hardcoded URL

1. Use WebFetch to call: `https://agentskill.sh/api/agent/search?q=<URL-encoded query>&limit=5`

Faible

SKILL.md:68

Hardcoded URL

If no results are found, say: "No skills found for '<query>'. Try different keywords or browse at ht

Faible

SKILL.md:77

Hardcoded URL

2. Use WebFetch to call: `https://agentskill.sh/api/agent/skills/<slug>/install?owner=<owner>`

Faible

SKILL.md:79

Hardcoded URL

4. If not found, say: "Skill @<owner>/<slug> not found. Check the name at https://agentskill.sh"

Faible

SKILL.md:87

Hardcoded URL

1. Parse the slug from the URL path (last segment of `https://agentskill.sh/<slug>`)

Faible

SKILL.md:88

Hardcoded URL

2. Use WebFetch to call: `https://agentskill.sh/api/agent/skills/<slug>/install`

Faible

SKILL.md:121

Hardcoded URL

1. Use WebFetch to call: `https://agentskill.sh/api/agent/search?section=trending&limit=5`

Faible

SKILL.md:133

Hardcoded URL

3. Use WebFetch to POST to `https://agentskill.sh/api/skills/<slug>/agent-feedback` with JSON body:

Faible

SKILL.md:179

Hardcoded URL

2. Collect all slugs and call the batch version endpoint: `https://agentskill.sh/api/agent/skills/ve

Faible

SKILL.md:256

Hardcoded URL

1. Fetch skill content from `https://agentskill.sh/api/agent/skills/<slug>/install?platform=<platfor

Faible

SKILL.md:325

Hardcoded URL

# source: https://agentskill.sh/<slug>

Faible

SKILL.md:331

Hardcoded URL

9. Track the install — use WebFetch to POST to `https://agentskill.sh/api/skills/<slug>/install` wit

Faible

SKILL.md:501

Hardcoded URL

POST https://agentskill.sh/api/agent/security-reports

Faible

SKILL.md:561

Hardcoded URL

2. Use WebFetch to call: `https://agentskill.sh/api/agent/skills/learn/version`

Faible

SKILL.md:565

Hardcoded URL

a. Fetch the latest version from `https://agentskill.sh/api/agent/skills/learn/install`

Faible

SKILL.md:768

Hardcoded URL

| API unreachable / timeout | "Could not reach agentskill.sh. Check your connection or try again lat

Faible

SKILL.md:769

Hardcoded URL

| No search results | "No skills found for '<query>'. Try different keywords or browse at ht

Faible

SKILL.md:770

Hardcoded URL

| Skill not found (404) | "Skill '<slug>' not found. It may have been removed. Browse available

Faible

SKILL.md:781

Hardcoded URL

All endpoints are on `https://agentskill.sh`.

Facteurs de risque

🌐 Accès réseau (30)

📁 Accès au système de fichiers (45)

⚡ Contient des scripts (3)

references/SECURITY.md:414 references/SECURITY.md:421 references/SECURITY.md:420

⚙️ Commandes externes (50)

references/SECURITY.md:412 references/SECURITY.md:413 references/SECURITY.md:411 references/SECURITY.md:630-635 references/SECURITY.md:149 SKILL.md:13 SKILL.md:22 SKILL.md:23 SKILL.md:24 SKILL.md:25 SKILL.md:32-34 SKILL.md:34-36 SKILL.md:36-40 SKILL.md:40-44 SKILL.md:44-57 SKILL.md:57-59 SKILL.md:59-70 SKILL.md:70-72 SKILL.md:72-76 SKILL.md:76-77 SKILL.md:77-81 SKILL.md:81-83 SKILL.md:83-87 SKILL.md:87-88 SKILL.md:88-91 SKILL.md:91-93 SKILL.md:93-98 SKILL.md:98-99 SKILL.md:99 SKILL.md:100 SKILL.md:101-103 SKILL.md:103 SKILL.md:104-105 SKILL.md:105-109 SKILL.md:109-113 SKILL.md:113-117 SKILL.md:117-121 SKILL.md:121-125 SKILL.md:125-131 SKILL.md:131 SKILL.md:133-134 SKILL.md:134-141 SKILL.md:141-144 SKILL.md:144-151 SKILL.md:151-153 SKILL.md:153-158 SKILL.md:158-159 SKILL.md:159 SKILL.md:162-171 SKILL.md:171-172

🔑 Variables d’environnement (1)

references/SECURITY.md:111

Motifs détectés

System reconnaissance[HEURISTIC] DANGEROUS COMBINATION: Code execution + Network + Credential access[HEURISTIC] DANGEROUS COMBINATION: Network + Credentials + Evasion techniques

Audité par: claude Voir l’historique des audits →

Partager et citer ce rapport

Partagez le rapport d’évaluation versionné, le badge neutre, la carte intégrable et les citations. Skillstore présente les preuves sans décider si ce Skill est sûr.

Ouvrir le rapport versionné

Copier le lien du rapport

https://skillstore.io/skills/agentskill-sh-learn/audits/7?utm_source=security_passport&utm_medium=share&utm_campaign=versioned_report

Badge Markdown

[![Skillstore security assessment](https://skillstore.io/badges/skills/agentskill-sh-learn/security.svg)](https://skillstore.io/skills/agentskill-sh-learn?utm_source=security_passport_badge)

Badge HTML

<a href="https://skillstore.io/skills/agentskill-sh-learn?utm_source=security_passport_badge"><img src="https://skillstore.io/badges/skills/agentskill-sh-learn/security.svg" alt="Skillstore security assessment" loading="lazy"></a>

Carte intégrable

<iframe src="https://skillstore.io/embed/skills/agentskill-sh-learn.html" title="Skillstore Security Assessment" sandbox="allow-popups allow-popups-to-escape-sandbox" loading="lazy" referrerpolicy="no-referrer" width="420" height="180"></iframe>

Citations académiques (APA · BibTeX · CFF)

Citation APA

agentskill-sh. (2026). learn security audit report (audit version 7) [Author version 2.0]. Skillstore. https://skillstore.io/skills/agentskill-sh-learn/audits/7

Citation BibTeX

@techreport{agentskill-sh-agentskill-sh-learn-2026,
  author = {agentskill-sh},
  title = {learn security audit report (audit version 7)},
  institution = {Skillstore},
  year = {2026},
  number = {7},
  url = {https://skillstore.io/skills/agentskill-sh-learn/audits/7},
  note = {Author version 2.0}
}

CITATION.cff

cff-version: 1.2.0
message: "If you use this Skill, cite its author and this versioned security audit report."
title: "learn security audit report (audit version 7)"
version: "2.0"
type: report
authors:
  - name: "agentskill-sh"
date-released: "2026-07-09"
url: "https://skillstore.io/skills/agentskill-sh-learn/audits/7"
identifiers:
  - type: other
    value: "skillstore:agentskill-sh-learn:audit:7"
    description: "Skillstore immutable audit report identifier"

Comparer les variantes

2 variantes installables

Chaque auteur reste un skill installable distinct. La variante recommandée est classée selon les preuves Skillstore.

Pourquoi cette variante est première

Meilleur score Skillstore

Auteur / source Score Skillstore Niveau de confiance des preuves Utilisation Skillstore Mis à jour

0xrdan Recommandé

0xrdan-learn

Score Skillstore 70

Niveau de confiance des preuves Élevé

Utilisation Skillstore 33

Mis à jour

2026-10-12

agentskill-sh Actuelle

agentskill-sh-learn

Score Skillstore 38

Niveau de confiance des preuves Moyen

Utilisation Skillstore 83

Mis à jour

2026-10-22

Score Skillstore

Pourquoi ce score Niveau de confiance des preuves: Moyen

Architecture

Maintenabilité

Contenu

Communauté

Conformité aux spécifications

Ce que vous pouvez créer

Trouver une capacité manquante

Rechercher une compétence correspondant à une tâche actuelle, comme le SEO, les paiements, la documentation ou les conseils sur un framework.

Gérer les compétences installées

Lister les compétences installées, vérifier les mises à jour, supprimer les compétences inutilisées et garder les métadonnées locales à jour.

Examiner la sécurité d'une compétence avant l'installation

Analyser une compétence locale ou distante à la recherche de motifs risqués avant de lui permettre d'influencer le comportement de l'agent.

Essayez ces prompts

Rechercher une compétence

/learn frontend react components

Installer une compétence connue

/learn @owner/skill-slug

Auditer une compétence locale

/learn scan ./SKILL.md

Mettre à jour les compétences gérées

/learn update

Bonnes pratiques

Examinez la prévisualisation, la source et le score de sécurité avant d'installer une compétence.
Utilisez le mode d'analyse pour les compétences locales ou tierces avant de les ajouter à un espace de travail d'agent.
Désactivez l'évaluation automatique si votre équipe ne souhaite pas envoyer de retour à un service externe.

Éviter

N'installez pas de compétences à faible score simplement parce qu'elles correspondent aux mots-clés de la tâche.
Ne considérez pas les analyses fondées sur des motifs comme un substitut complet à l'examen manuel d'instructions non fiables.
N'autorisez pas les mises à jour de compétences distantes dans des espaces de travail sensibles sans examiner le contenu modifié.

Foire aux questions

Que fait cette compétence ?

Elle recherche, prévisualise, installe, met à jour, supprime, analyse et évalue les compétences d'agents IA depuis agentskill.sh.

Fonctionne-t-elle avec Claude Code ?

Oui. La compétence déclare prendre en charge Claude, Codex et Claude Code.

Installe-t-elle du code automatiquement ?

Elle peut écrire des fichiers de compétences après une étape de prévisualisation, d'analyse de sécurité et de confirmation utilisateur décrite dans les instructions.

L'analyse de sécurité peut-elle prouver qu'une compétence est sûre ?

Non. Elle détecte des motifs risqués connus, mais les utilisateurs doivent tout de même examiner le contenu des compétences non fiables.

Quelles données peut-elle envoyer à l'extérieur ?

Elle contacte agentskill.sh pour les flux de travail de recherche, d'installation, de mise à jour, de suivi des installations, de retour et de rapport de sécurité.

Quand les équipes doivent-elles l'éviter ?

Évitez-la dans les environnements restreints où les instructions distantes, la télémétrie ou les écritures de compétences dans l'espace de travail ne sont pas autorisées.