Back to skill
Skillv1.0.0
ClawScan security
Replication Strategy Selector · ClawHub's context-aware review of the artifact, metadata, and declared behavior.
Scanner verdict
BenignApr 11, 2026, 11:31 AM
- Verdict
- benign
- Confidence
- high
- Model
- gpt-5-mini
- Summary
- The skill's requirements and instructions are coherent with its stated purpose: it is an instruction-only helper for selecting and configuring replication strategies and does not request unrelated credentials or install code.
- Guidance
- This skill appears coherent and focused on advising/configuring replication. Before installing or running it, consider: (1) it expects access to your project directory and config files — remove or redact any secrets (API keys, DB passwords) from those files or run the skill in a sanitized copy; (2) the execution metadata requests Bash and Write, so the agent could modify files — prefer running it in a branch or sandbox and review any changes before applying them to production; (3) because the skill can inspect config and code, limit its runtime permissions (filesystem scope, network access) to the minimum necessary; (4) if you need the agent to only produce recommendations without editing files, restrict or disable the Write/Bash tools at runtime. Overall the skill is internally consistent with its stated purpose.
- Findings
[regex-scan-none] expected: The regex-based scanner had nothing to analyze because this is an instruction-only skill composed of markdown references. Absence of findings is expected here and does not imply absence of risk at runtime (the agent may still be granted shell/write access).
Review Dimensions
- Purpose & Capability
- okThe name and description match the actions the skill asks the agent to perform: inspect an application codebase or architecture doc and recommend topology, sync mode, quorum values, and conflict resolution. There are no unexpected environment variables, external services, or binaries required that would be disproportionate to a replication-advice skill. The declared dependency on related selector/analyzer skills is plausible.
- Instruction Scope
- okThe SKILL.md instructs the agent to gather context from project files, config files, or system requirements documents — exactly the inputs you'd expect for choosing replication strategy. It does not instruct the agent to read unrelated system paths, secrets, or external endpoints. The presence of reference documents included with the skill is consistent with an advisory role.
- Install Mechanism
- okThere is no install spec and no code files to run; this is an instruction-only skill. That minimizes the risk of arbitrary code being downloaded or executed by the skill itself.
- Credentials
- okThe skill does not request environment variables, credentials, or config paths. The inputs it asks for (codebase, config files, architecture docs) are proportionate to producing replication recommendations.
- Persistence & Privilege
- notealways:false (normal). Execution metadata requires tools: Read, Write, Bash. This is coherent because the skill may need to read configs and optionally apply or patch config files, but it does grant the agent the ability to run shell commands and write files in the project directory — a powerful capability that the user should control via run-time permissions and review of proposed changes.