lgd-skill-review@1.1.21
VERIFIED· install onlyPublisher claims: "Use when reviewing, auditing or scoring any AI skill package, or when a skill was rejected and you need to know why. A dual-ruler reviewer: reports a structure score (100) and a semantics score (120) as two orthogonal verdicts instead of merging them, and flags the case when the two disagree by 20 points or more. Structure asks whether the package can be used at all (tests, declared-vs-implemented consistency, reproducibility, hygiene, discovery, licensing); semantics asks whether it is worth using (knowledge delta, procedures, anti-patterns with reasons, specification, progressive disclosure, freedom calibration, pattern fit, usability). Every finding carries a stable ID and is traceable to a specific file. Accepts evidence-based appeals: the appellant supplies a path and a check, the engine opens it and rules OVERTURNED / UPHELD, and records the ruling in appeal_log.json. Appeals never silently change a score - they turn a deduction into a verifiable dispute." Observed in sandbox: no network, touches nothing outside the project, writes no memory.
Coverage
Bundled scripts ran with no arguments; no agent task scenario for this bundle.- Tested
- 3 of 3 bundled scripts run with no arguments, with decoy credentials in the environment · every file read, file written and host contacted by what ran
- Not tested
- use of the skill by an agent
What changed since 1.1.20
vet-skills diff lgd-skill-review 1.1.20 1.1.21In the sandbox, this version reached for nothing 1.1.20 did not. Compared from the two recorded runs, not from source.
The guard can hold an update like this until someone approves it (runtime.on_change: hold). A change that only shows with a real credential, or on a path the run did not reach, is not in a trace and so not in this comparison.
Is [email protected] safe to install?
KillerSkills ran it in the KillerSkills honeypot sandbox (2 scenarios; bundled scripts ran with no arguments; no agent task scenario for this bundle) on 2026-11-18 and scored it 88/100 (Verified). In what ran, nothing we could detect went wrong: no undeclared network egress, no credential reads, no persistent instruction writes. What it does with a real credential was not observed; the static findings and the publisher carry the rest of the score. Summary: no network, touches nothing outside the project, writes no memory.
Behavior trace
Honeypot run · 0 s · 2 scenarios · 0 deviationsStatic findings
p = subprocess.run([sys.executable, REVIEW, root, "--run"],Fix: Build commands from fixed arguments, never from interpolated strings; prefer an API client over shelling out.…ine opens it and rules OVERTURNED / UPHELD, and records the ruling in appeal_log.json. Appeals never silently change a score - they turn a deduction into a verifiable dispute.' license: MIT allowed-tools: Read, Glob, Grep, Bash metadata: d…Fix: Remove the call, or declare the host in frontmatter and explain in the README what is sent and why.for pat in ["eval(", "exec(", "os.system", "pickle.loads"]:Fix: Put hosts, commands and strings in plain source. Decoding at runtime looks identical to malware and costs points on its own.r = subprocess.run([sys.executable, runner], cwd=root, capture_output=True,Fix: Build commands from fixed arguments, never from interpolated strings; prefer an API client over shelling out.Cost to the agent
Tokens estimated at four characters each. Skills load their instruction text when triggered; the figure is what that costs.
Known vulnerabilities
Vulnerability advisories (OSV) cover npm and PyPI packages; this registry has no advisory database.