Skillv2.1.0

ClawScan security

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

BenignApr 27, 2026, 3:21 AM

Verdict: benign
Confidence: high
Model: gpt-5-mini
Summary: The skill's code and instructions match its stated purpose (local token/cost reporting) and do not request extra credentials, network access, or unusual installs.
Guidance: This skill appears to do exactly what it claims: read session metadata (via session_status and sessions_list), run a local Python script, and produce a token/cost report. Before installing, confirm you are comfortable the platform APIs (session_status/sessions_list) expose only the session metadata you expect (tokens, model, timestamps) because the report includes session identifiers and usage totals. You can also review scripts/cost.py in your environment (it's plain Python, no networking or subprocess calls) to verify no hidden behavior. If you use shared or sensitive sessions, note that top-sessions output may reveal session IDs and usage patterns in the chat output.

Purpose & Capability: okName, description, SKILL.md and scripts/cost.py align: the skill reads session_status and sessions_list, computes token costs and projections, and prints a report. No unrelated credentials, binaries, or install steps are required.
Instruction Scope: okRuntime instructions are limited to calling the platform-provided APIs (session_status, sessions_list) and invoking the bundled scripts/cost.py with calculated arguments. The instructions do not direct reading of arbitrary files, accessing unrelated env vars, or sending data to external endpoints. Note: the output may include session IDs (shortened) and cost/credit figures derived from the session data.
Install Mechanism: okThis is an instruction-only skill with a local Python script; there is no install spec, no downloads, and no third-party package installation. Risk from install mechanism is minimal.
Credentials: okThe skill declares no required environment variables or credentials and the code does not read environment secrets. The inputs are provided by the runtime (session data) and command-line parameters; requested data (tokens, model, credits) are proportionate to computing costs.
Persistence & Privilege: okalways is false and the skill does not request persistent system presence or modify other skills. It simply runs a local script when invoked.