Claude Code, Skill
Skill Quality Research
Review any Agent Skill repo for how well it is WRITTEN, not just how well it packages.
Compatibility: Not stated.
Grade B
Minor issues only, no Medium or High. Worth a quick look at the findings below before installing.
Not a guarantee. A hygiene check of the source at the commit shown, not a security audit.
NETWORK_NO_FIXED_HOST SECRETS_PATH_DOTENV PERMISSION_SHELL_DECLARED INSTALL_DOCUMENTED_INSTALLER SECRETS_ENV_READ OBFUSCATION_FILE_OVER_READ_CAP
Why this grade
- NETWORK_NO_FIXED_HOST Low
Makes a network call whose host is not written as a fixed address.
scripts/backfill_repo_fields.py, line 19
def fetch(full_name): - SECRETS_PATH_DOTENV Low
Mentions a .env file.
CHANGELOG.md, line 389
的反向實錯(審查者看到 `.env` 就斷定誤報,查證後該 CLI 真的實作了 `--api_key`), - PERMISSION_SHELL_DECLARED Note
Runs shell commands, and the manifest or description says so.
scripts/backfill_repo_fields.py, line 20
r = subprocess.run( - INSTALL_DOCUMENTED_INSTALLER Note
Documentation shows an installer that pipes a download into a shell.
research/blind-craft-reviews-2026-08-27/NevaMind-AI__memU.md, line 166
> `- macOS / Linux: `curl -fsSL https://claude.ai/install.sh | bash`` - SECRETS_ENV_READ Note
Reads an environment variable whose name contains KEY, SECRET, TOKEN, PASSWORD or CREDENTIAL.
scripts/collect_repos.py, line 117
tok = os.environ.get("GITHUB_TOKEN") or os.environ.get("GH_TOKEN") - OBFUSCATION_FILE_OVER_READ_CAP Note
A file was over the read cap and was not fully read.
research/feature_matrix.json, line 1
{
Any High, or two or more Medium findings, give D. One Medium gives C. Low findings or a security Note give B. Each code names a finding behind the letter.
Grades are a hygiene check of the source at the commit shown, on the date checked, not a security audit or a guarantee. The free directory grades are automated. Before release, a PlugVet engineer reviews every paid report and any grade we hold back, including all F grades, against the source code and the evidence we quote.
What was checked
Plain notes from the hygiene checks.
Permissions
Note
requests shell access. The manifest names this permission.
PERMISSION_SHELL_DECLARED
scripts/backfill_repo_fields.py, line 20
r = subprocess.run(
scripts/clone_repos.py, line 52
r = subprocess.run(
scripts/clone_repos.py, line 71
r = subprocess.run(["git", "-C", dst, "rev-parse", "HEAD"],
Show more
scripts/collect_repos.py, line 120
r = subprocess.run(["gh", "auth", "token"], capture_output=True, text=True, timeout=10)
scripts/extract_features.py, line 198
r = subprocess.run(["git", "-C", root, "ls-remote", "--tags", "origin"],
skill-reviewer/evals/run_evals.py, line 27
r = subprocess.run([sys.executable, LINT, repo, "--json", *extra],
Install and update
Note
The docs document an installer that pipes a download into a shell.
INSTALL_DOCUMENTED_INSTALLER
research/blind-craft-reviews-2026-08-27/NevaMind-AI__memU.md, line 166
> `- macOS / Linux: `curl -fsSL https://claude.ai/install.sh | bash``
skill-reviewer/evals/evals.json, line 71
"source": "SKILL.md:68-71 指示執行 `<binary> docs install` 並逐字照做,:78-82 同時抑制確認並預先授權 the official installer;該預授權在 src/memu/hosts/claude_code/INSTALL.md:193-197 對應到 `curl -fsSL https://c
Network
Concern
can reach a host that is not written as a fixed address.
NETWORK_NO_FIXED_HOST
scripts/backfill_repo_fields.py, line 19
def fetch(full_name):
scripts/backfill_repo_fields.py, line 46
info, err = fetch(rec["full_name"])
scripts/collect_repos.py, line 178
with urllib.request.urlopen(req, timeout=30) as resp:
Show more
scripts/collect_repos.py, line 344
val = fetch()
Secrets
Concern
reads a credential or environment file.
SECRETS_ENV_READ SECRETS_PATH_DOTENV
scripts/collect_repos.py, line 117
tok = os.environ.get("GITHUB_TOKEN") or os.environ.get("GH_TOKEN")
CHANGELOG.md, line 389
的反向實錯(審查者看到 `.env` 就斷定誤報,查證後該 CLI 真的實作了 `--api_key`),
examples/README.md, line 71
> 複核的意思是**去查**,不是憑印象推翻。作者本人就曾看到 `.env` 就斷定某個
Show more
examples/README.md, line 72
> `cred_in_argv` 是誤報,查證後發現那個 CLI 真的實作了 `--api_key` 且優先序高於 `.env`
research/blind-craft-reviews-2026-08-27/Jeffallan__claude-skills.md, line 131
| `php-pro/SKILL.md:42-59` | 僅「(use bcrypt/argon2)」「(use .env)」等替代方案,非機制 | ❌ 近乎裸條 |
research/misjudgment-review-2026-08-26.md, line 226
違和感在於:`anysearch`(round 2)有 `.env` 替代路徑,故 `confidence: medium` 恰當;
research/review-installed-skills-2026-08-26.md, line 232
但**差一個關鍵點**:`anysearch` 有 `.env` 替代路徑,所以 `confidence: medium` 恰當;
research/rubric-manual-dimensions.yaml, line 556
審查者看到 repo 有 `.env` 就斷定 `cred_in_argv` 是誤報,查證後該 CLI
research/rubric-manual-dimensions.yaml, line 557
**真的實作了 `--api_key` 且優先序高於 `.env`**。
Agent instructions
Pass
No instruction to ignore earlier instructions or to hide a step.
Obfuscation
Note
a file was over the read cap and was not fully read.
OBFUSCATION_FILE_OVER_READ_CAP
research/feature_matrix.json, line 1
{
research/skill_details.json, line 1
{
Licence
Pass
A licence file is present (LICENSE).
Maintainer activity
Pass
The latest commit is within the last 365 days.
Note
Hygiene check of commit 62da064c168f. requests shell access. The manifest names this permission. The docs document an installer that pipes a download into a shell. can reach a host that is not written as a fixed address. reads a credential or environment file. a file was over the read cap and was not fully read.
Date and source
Date checked:
github.com
Commit graded: 62da064c168f391b91e372a84166bc682054f909
This link goes to the original host. PlugVet does not host or check what you download from there, and installing it is your own decision and risk.