ClawHub

评论透镜

v1.0.0

Review intelligence skill for mainland China shopping that compresses large volumes of marketplace reviews into a decision card, surfaces repeated praise and...

⭐ 0· 32·0 current·0 all-time
byhaidong@harrylabsj
MIT-0
Download zip
LicenseMIT-0 · Free to use, modify, and redistribute. No attribution required.
Security Scan
VirusTotalVirusTotal
Benign
View report →
OpenClawOpenClaw
Benign
high confidence
✓
Purpose & Capability
Name/description match the included SKILL.md and reference docs: the skill is a review-intelligence layer that summarizes review patterns, clusters complaints/praise, and issues a buying recommendation. Required env, binaries, and config paths are empty — appropriate for an instruction-only review analysis skill.
✓
Instruction Scope
SKILL.md confines the agent to reading user-provided input (links, screenshots, copied review text, rating counts) and applying clustering/heuristics from the references. It does not instruct the agent to read unrelated local files, access system secrets, or call unexpected external endpoints. The guidance is prescriptive rather than open-ended.
✓
Install Mechanism
No install spec; only a publish helper script (scripts/publish.sh) is included for maintainers. The script merely packages files and calls clawhub publish (and requires node/clawhub for publishing), which is normal and not executed at runtime for end users.
✓
Credentials
The skill declares no required environment variables, no primary credential, and no config paths. There are no unexpected credential requests and the scope of access matches the task of processing user-supplied reviews.
✓
Persistence & Privilege
always is false and disable-model-invocation is false (normal). The skill does not request persistent system presence or altered agent/global config; nothing suggests elevated privileges.
Assessment
This skill is internally coherent and appears to do what it says: summarize and judge product reviews. Before installing, consider these practical points: (1) The skill expects you to supply review text, links, or screenshots — avoid uploading images or screenshots that contain other people's personal data or account tokens. (2) The included publish script is only relevant to the author/maintainer; it is not an install-time action for end users. (3) If you plan to have the agent follow product links automatically, confirm whether your agent runtime will fetch those URLs (network fetches are separate from this skill and you should review agent network policies). (4) As always, test with non-sensitive sample data to validate results and watch for hallucinated claims about review counts or timelines.

Like a lobster shell, security has layers — review code before you run it.

buyer-regretvk979fwyd8pq6mh9djgdazex7c1847g0secommercevk979fwyd8pq6mh9djgdazex7c1847g0sjdvk979fwyd8pq6mh9djgdazex7c1847g0slatestvk979fwyd8pq6mh9djgdazex7c1847g0spddvk979fwyd8pq6mh9djgdazex7c1847g0sreview-analysisvk979fwyd8pq6mh9djgdazex7c1847g0sreview-intelligencevk979fwyd8pq6mh9djgdazex7c1847g0sreviewsvk979fwyd8pq6mh9djgdazex7c1847g0sshoppingvk979fwyd8pq6mh9djgdazex7c1847g0staobaovk979fwyd8pq6mh9djgdazex7c1847g0stmallvk979fwyd8pq6mh9djgdazex7c1847g0suser-fitvk979fwyd8pq6mh9djgdazex7c1847g0s

License

MIT-0
Free to use, modify, and redistribute. No attribution required.
Termshttps://spdx.org/licenses/MIT-0.html

Runtime requirements

🔍 Clawdis
OSLinux · macOS · Windows

Comments