OpenAgentSkill
GitHub200+

Evaluasi sebelum pemasangan

eval-design Laporan evaluasi.

Keputusan pemasangan yang dapat dibaca mesin untuk Agent: kecocokan tugas, Trust Score, Audit Score, keamanan pemasangan, cakupan izin, dan rencana validasi konkret sebelum skill menyentuh workspace.

Perlu ditinjauRisiko sedangTinjau Kebijakan
70
Laporan evaluasi
73
Kepercayaan
77
Audit
45
Keamanan Agent

Tinjauan manual

Test manually in an isolated workspace and compare against safer alternatives.

Gerbang wajib

Pemeriksaan yang harus dilalui Agent sebelum memasang

Buka JSON

Kecocokan tugas

94

Lulus

Task wording matches this skill metadata.

  • Evaluate eval-design before installing it in an agent workflow
  • Riset
  • Web scraping workflows; Claude Code teams; teams that value GitHub adoption signals

Jalur pemasangan

92

Lulus

Install handoff is available.

  • npx skills add agentscope-ai/OpenJudge --skill eval-design

Keamanan perintah pemasangan

68

Peringatan

dynamic command execution, standard package or runtime install path

  • npx skills add agentscope-ai/OpenJudge --skill eval-design

Skor kepercayaan

73

Peringatan

Good trust signals with a few areas worth checking before rollout.

  • Shortlist kuat
  • 816 star GitHub
  • Apache-2.0

Skor audit

77

Peringatan

Perlu ditinjau

  • Minor inconsistency: SKILL.md states a 60/30/10 difficulty split, but the bundled coverage_check.py uses EXPECTED_STRATA = {'easy': 0.55, 'boundary': 0.30, 'adversarial': 0.10}. This could confuse users relying on the documented split.

Gerbang keamanan Agent

45

Peringatan

Sparse or mixed signals. Useful for discovery, but not for autonomous installation.

  • Test manually in an isolated workspace and compare against safer alternatives.
  • Petunjuk izin berisiko tinggi: eksekusi shell atau perintah

Kejelasan lisensi

86

Lulus

Apache-2.0

  • Apache-2.0

Cakupan izin

62

Peringatan

shell or command execution, filesystem or document access

  • Shell or command execution: high
  • Network access: medium
  • Filesystem access: medium

Rencana validasi

Langkah Agent berikutnya

  1. 1Inspect repository, README/SKILL.md, license, and recent commits before production use.
  2. 2Install in an isolated workspace or sandbox with no production secrets available.
  3. 3Run the smallest representative task and record files touched, commands run, network access, and outputs.
  4. 4Compare the selected skill against at least one alternative when the eval status is review or failed.
  5. 5Promote only after the agent reports a successful verification result and unresolved warnings are accepted.

Jangan gunakan ketika

Kondisi yang membutuhkan skill lain

  • Tim yang membutuhkan SLA dengan dukungan vendor
  • production agents without a repository review
  • Minor inconsistency: SKILL.md states a 60/30/10 difficulty split, but the bundled coverage_check.py uses EXPECTED_STRATA = {'easy': 0.55, 'boundary': 0.30, 'adversarial': 0.10}. This could confuse users relying on the documented split.
  • Petunjuk izin berisiko tinggi: eksekusi shell atau perintah
  • Quality score needs review
  • Production credentials, payments, or irreversible account changes without explicit human review

Pemeriksaan pendukung

Sinyal kepercayaan di balik keputusan

Kelengkapan README/SKILL.md

Lulus

86

Metadata memuat konteks penggunaan dan alur kerja yang cukup

Pemeliharaan terbaru

Lulus

88

2 bulan sejak push

Alternatif tersedia

Lulus

82

Alternative skills are available for comparison.