Skillv1.0.0

ClawScan security

SpecVibe · ClawHub's context-aware review of the artifact, metadata, and declared behavior.

Scanner verdict

BenignFeb 25, 2026, 8:58 AM

Verdict: benign
Confidence: high
Model: gpt-5-mini
Summary: The skill's instructions, templates, and references are coherent with its stated purpose as a spec-driven development framework and do not request unrelated credentials or installs.
Guidance: This package appears internally consistent with its stated goal: it is an instruction-only framework with templates and reference docs and asks for no secrets or remote installs. Before installing, verify the origin (the 'Source' is unknown and README links to a GitHub path), and inspect the files locally. If you install by copying into an assistant skill directory, understand that those files will influence the assistant's behavior persistently — only proceed if you trust the author. Do not give the AI autonomous permission to push commits, run CI, or access secret stores without human review; require manual approval for actions that modify your repository, environment variables, or deployment pipelines.

Review Dimensions

Purpose & Capability: okThe name/description (a spec-driven development framework) matches the included SKILL.md, reference guides, and templates. The files provided (spec, openapi template, CI/devops/security guidance, cursorrules template) are exactly what a framework of this type would include.
Instruction Scope: noteSKILL.md instructs the AI and the user to generate spec.md, PLAN.md, tests, code, CI, and to integrate the repository into assistant skill directories (e.g., ~/.openclaw/skills/, .claude/skills/). Those actions are within the framework's purpose, but they do imply writing files to your environment and having the AI operate over your repo (e.g., using context-packaging tools like gitingest/repo2txt). There are no instructions to read unrelated secrets or system files, but installing into an assistant's skill directory will make these rules active for that assistant.
Install Mechanism: okNo install spec or remote downloads are present; this is an instruction-only skill with all content included. That lowers risk compared to remote installs.
Credentials: okThe skill declares no required environment variables, binaries, or credentials. The references recommend standard practices (use env vars and secret managers) but do not request any unrelated secrets or access tokens.
Persistence & Privilege: noteThe README explicitly suggests cloning the repo into assistant skill directories so the assistant will load it. The skill metadata itself does not set always:true. Be aware that placing these files into an assistant's skill folder gives them persistent influence over the assistant's behavior; this is expected for a framework but is a persistence point you should consciously approve.