Bookmarks
v1.0.0Build a unified bookmark system that imports saves from all your platforms into one organized, actionable collection.
⭐ 3· 985·1 current·1 all-time
byIván@ivangdavila
MIT-0
Download zip
LicenseMIT-0 · Free to use, modify, and redistribute. No attribution required.
Security Scan
OpenClaw
Suspicious
high confidencePurpose & Capability
The skill's purpose is to connect to external platforms (X, YouTube, Reddit, Pinterest, Instagram, TikTok) and import saves. However the registry metadata declares no required credentials, no primary credential, no config paths, and no install. Connecting to those services normally requires OAuth tokens or API keys — the skill provides no mechanism or declared requirement for obtaining or storing them. That is an incoherence: a bookmark importer legitimately needs access credentials or an OAuth flow but none are specified.
Instruction Scope
SKILL.md instructs the agent to 'import saves — silently, no questions', 'works silently in background', and to create and maintain files under ~/bookmarks/. It lacks any detail on how to obtain user authorization, does not constrain what is imported beyond a short 'explicit saves only' note, and gives the agent broad discretion (silent imports, auto-tagging, periodic summaries). Those runtime directives expand the agent's scope beyond what's declared and omit essential consent/authorization steps.
Install Mechanism
There is no install spec and no code files — that reduces supply-chain risk because nothing is downloaded or written by an installer. However, the skill claims ongoing background behavior and periodic actions but gives no mechanism (no daemon, no scheduled job, no persistent component) to achieve that. That mismatch is an operational incoherence: background operation is claimed but not enabled by any install/runtime instructions.
Credentials
The skill declares zero required environment variables or credentials despite describing functionality that normally requires access tokens or OAuth flows to third-party platforms. Required secrets are absent from the spec and not discussed in SKILL.md. This omission either (a) hides where credentials would come from, or (b) assumes the agent will access user accounts by other means — both are concerning and disproportionate to the clear stated purpose.
Persistence & Privilege
always:false (good) and model invocation is allowed (normal). But the skill's narrative of 'silent background' operation and periodic summaries implies persistent or scheduled activity. Without an install/runtime mechanism or declared scheduling permissions, it's unclear how persistence would be achieved. If the skill were later granted autonomous periodic invocation, its silent import behavior raises privacy risks — combine that with the missing credential spec and you get elevated concern.
What to consider before installing
This skill has clear gaps between what it says it will do and what it actually declares it needs. Before installing or enabling it, ask the publisher: (1) How will the skill obtain access to third-party accounts (OAuth flows, where are tokens stored, and who can read them)? (2) Where and with what file permissions will ~/bookmarks/ be created, and will data be encrypted at rest? (3) Will imports run automatically or only when you explicitly approve them? The SKILL.md currently instructs silent, automatic imports — request an explicit consent workflow and a minimal-scope OAuth design. Also ask for an install/runtime plan that explains how background/periodic behavior is implemented (if at all). If the publisher cannot clearly explain credential handling, consent prompts, and persistence mechanics, do not enable the skill with access to your accounts or sensitive data. Consider testing in a sandboxed account or rejecting the skill until these gaps are remedied.Like a lobster shell, security has layers — review code before you run it.
latestvk97a0a64nx6ngdqsr0vrbwr69s80y6a5
License
MIT-0
Free to use, modify, and redistribute. No attribution required.
Runtime requirements
🔖 Clawdis
OSLinux · macOS · Windows