Back to skill
Skillv0.1.0
ClawScan security
Claude Cost Cli · ClawHub's context-aware review of the artifact, metadata, and declared behavior.
Scanner verdict
BenignFeb 11, 2026, 7:28 PM
- Verdict
- benign
- Confidence
- high
- Model
- gpt-5-mini
- Summary
- The skill's required binaries, storage method, and runtime instructions are consistent with a macOS CLI that queries Anthropic/Claude Admin API usage and costs; nothing requested or instructed appears disproportionate to that purpose.
- Guidance
- This skill appears internally consistent for a macOS CLI that queries Anthropic/Claude usage and costs. Before installing: 1) Verify the npm package and GitHub repository (check the provenance attestation or review the source code) to reduce supply-chain risk; 2) Confirm the Admin API key you provide is the intended read-only Admin usage key and understand who can generate/revoke it in your organization; 3) Prefer installing from audited source (git clone + npm build) if you need higher assurance; 4) Remember the CLI will store the key in macOS Keychain — review that entry if you later revoke access; 5) Because the skill runs an external binary (claude-cost), only install/run it on machines you trust. If you need greater assurance, ask the publisher for a signed release or an audit of the repository.
Review Dimensions
- Purpose & Capability
- okName/description (Claude cost/usage CLI) match the artifacts: it requires the 'claude-cost' CLI and node, documents using an Anthropic Admin API key, and describes Keychain storage and usage/cost endpoints — these requirements are expected for this purpose.
- Instruction Scope
- okSKILL.md only instructs installation of the CLI (npm or from source), running the 'claude-cost' commands, and storing the Admin API key in macOS Keychain. It does not ask the agent to read unrelated files, access other credentials, or transmit data to endpoints outside api.anthropic.com per the documented behavior.
- Install Mechanism
- noteThis is an instruction-only skill (no code bundled). The README suggests installing via npm (npm install -g claude-cost-cli) or building from the GitHub repo — installing from npm is a common, moderate-risk supply-chain operation. Registry metadata shows no enforced install spec, but SKILL.md includes an npm install path and a GitHub source link; verify the package provenance and source before installing.
- Credentials
- okNo environment variables or unrelated credentials are requested. The only secret required is an Anthropic Admin API key (stored in macOS Keychain), which is proportional to querying organization usage/cost data.
- Persistence & Privilege
- okThe skill does not request persistent 'always' inclusion, does not modify other skills or system-wide agent settings, and limits storage to a Keychain entry (per the docs). Autonomous model invocation is allowed by default but is not combined with broad privileges here.