Skillv1.0.0

ClawScan security

N Plus One Detector · ClawHub's context-aware review of the artifact, metadata, and declared behavior.

Scanner verdict

BenignApr 29, 2026, 11:35 PM

Verdict: benign
Confidence: high
Model: gpt-5-mini
Summary: The skill's instructions and requirements are coherent with its purpose (finding N+1 queries); it doesn't request extra credentials or installs, but it expects you to run code/searches locally and enable query logging which can execute code and surface sensitive data.
Guidance: This skill is coherent for finding N+1 query problems, but before running its commands: 1) run them in a local/dev or CI test environment — do not enable query logging or execute tests against production databases, since logs can reveal sensitive data; 2) ensure you have the required tooling (ripgrep 'rg', python3, node/prisma tooling) available in the environment; 3) review any code snippets that the skill suggests running (they may execute your app code) and prefer read-only/logging modes where possible; 4) if you need to run runtime detection, use a sandboxed environment or test DB snapshot to avoid side effects. If you want, I can list the exact commands the skill will run and highlight which ones execute application code vs. just scan files.
Findings: [no_code_files_to_scan] expected: The skill is instruction-only (only SKILL.md present), so the regex scanner had nothing to analyze. Absence of findings is expected and not evidence of safety.

Review Dimensions

Purpose & Capability: noteThe name/description match the instructions: the SKILL.md contains targeted code-search patterns, examples for Django/Prisma/Rails/etc., and fixes (select_related, DataLoader, batching). One minor mismatch: the instructions assume availability of tools (rg, python3/node) and test/dev servers but the skill metadata does not declare required binaries — this is not malicious but the runtime environment must provide those tools.
Instruction Scope: noteInstructions stay on-topic (static grep searches, runtime query logging, generating reports, adding middleware/tests). However the runtime steps include running snippets against the application (enabling DB query logging, running views/tests, running code snippets) which will execute code in the target repo and may have side effects or surface sensitive DB contents in logs. This is expected for this purpose but should be run in a safe/dev environment.
Install Mechanism: okNo install spec and no code files — lowest risk from installation. The skill is instruction-only and writes nothing to disk by itself.
Credentials: okThe skill requests no environment variables or credentials. The SKILL.md suggests temporarily setting debug flags (e.g., DJANGO_DEBUG=1 or DEBUG=prisma:query) to gather query logs; these are appropriate for diagnosing N+1 issues and do not require persistent secrets.
Persistence & Privilege: okSkill is not always-enabled and does not request persistent privileges or modify other skills. It is user-invocable and can be used autonomously by the agent (default), which is normal for skills and not in itself concerning.