Skill flagged — suspicious patterns detected
ClawHub Security flagged this skill as suspicious. Review the scan results before using.
Daxiang Dogfood
v1.0.0Systematically explore and test a web application to find bugs, UX issues, and other problems. Use when asked to "dogfood", "QA", "exploratory test", "find i...
⭐ 0· 30·1 current·1 all-time
MIT-0
Download zip
LicenseMIT-0 · Free to use, modify, and redistribute. No attribution required.
Security Scan
OpenClaw
Benign
high confidencePurpose & Capability
Name/description (web app exploratory testing) match the instructions and bundled templates. The skill is instruction-only and relies on agent-browser to drive a browser and produce screenshots/videos/repro steps, which is expected for this purpose.
Instruction Scope
Instructions are explicit and scoped to testing: initializing a session, authenticating (if provided), navigating pages, taking snapshots, recording videos, saving state, and writing a report. This is within purpose, but the instructions direct the agent to capture potentially sensitive artifacts (screenshots, videos, saved auth-state JSON) and to save them to disk — the user should expect sensitive data to be stored in the output directory.
Install Mechanism
There is no install spec and no code files to execute; the skill is instruction-only. This minimizes install-time risk.
Credentials
The skill declares no required environment variables or credentials, but it explicitly instructs the agent to accept credentials when the user provides them and to save auth state. That behavior is appropriate for testing authenticated apps but means credentials/session tokens may be written to disk — users should only provide credentials they control and secure the output directory.
Persistence & Privilege
always is false, and the skill does not request elevated platform persistence or modify other skills. It only writes files within the output directory it creates.
Assessment
This skill appears coherent for QA dogfooding, but be cautious about what you test and where: only run it against apps you own or have explicit permission to test. The tool records screenshots, videos, and saved auth state (cookies/tokens) into the output directory — those artifacts can contain sensitive data, so secure or delete them after use. Prefer using throwaway/test accounts or ephemeral credentials for authenticated sessions. Note a small inconsistency: SKILL.md advises using the direct agent-browser binary (fast Rust client) but allowed-tools includes an npx variant; this is likely benign but you may want to confirm which binary the agent will actually invoke. If you do not want the agent to run autonomously against live hosts, avoid invoking the skill or restrict its use to manual runs.Like a lobster shell, security has layers — review code before you run it.
latestvk9700y4ct6bqck75zrd02shs3583ypsj
License
MIT-0
Free to use, modify, and redistribute. No attribution required.