Back to skill
Skillv1.0.2
ClawScan security
AI Search by Desearch · ClawHub's context-aware review of the artifact, metadata, and declared behavior.
Scanner verdict
BenignFeb 24, 2026, 5:31 PM
- Verdict
- Benign
- Confidence
- high
- Model
- gpt-5-mini
- Summary
- The skill's code, instructions, and required environment variable are coherent with a multi-source search CLI that sends queries to desearch.ai; nothing disproportionate or unexpected was found.
- Guidance
- This skill appears coherent and implements a small CLI that sends queries to desearch.ai using an API key. Before installing, verify the provider (console.desearch.ai / desearch.ai) and its privacy/terms so you understand how queries and results are handled. Treat DESEARCH_API_KEY like a secret — do not reuse other credentials as this key grants the skill network access. Because the agent can call the skill, avoid sending sensitive or private data through it unless you trust the service. You may also review/run the included scripts/desearch.py in a sandbox or inspect network traffic to confirm behavior before use.
Review Dimensions
- Purpose & Capability
- okName/description, SKILL.md, and the included Python CLI all align: the tool queries desearch.ai endpoints and returns aggregated search/summarization results. The single required env var (DESEARCH_API_KEY) is exactly what you'd expect for an external API.
- Instruction Scope
- okRuntime instructions only describe obtaining an API key, setting DESEARCH_API_KEY, and running the CLI. The script reads that env var and issues HTTP requests to https://api.desearch.ai; it does not read other files, execute shell commands, or contact unrelated endpoints.
- Install Mechanism
- okThere is no install spec and no downloads; the skill is instruction-only with one included Python script. No remote installs, archive extraction, or package pulls were specified.
- Credentials
- okOnly DESEARCH_API_KEY is required. That is proportional and appropriate for a service that calls a paid/authorized API. The code does not request other secrets or system config paths.
- Persistence & Privilege
- notealways:false and normal model invocation settings (agent may call the skill autonomously). This is the platform default and not inherently problematic, but remember the skill transmits whatever queries it receives to an external API.